SYSTEMATICS Tests of Current Hypotheses of Mayfly (Ephemeroptera) Phylogeny Using Molecular (18s rDNA) Data

LU SUN, ANIKO SABO,1 M. D. MEYER,2 R. P. RANDOLPH,3 LUKE M. JACOBUS, W. P. MCCAFFERTY, AND VIRGINIA R. FERRIS

Department of Entomology, Purdue University, West Lafayette, IN 47907

Ann. Entomol. Soc. Am. 99(2): 241Ð252 (2006) ABSTRACT Partial 18s rDNA sequences from 22 exemplar mayßy species (Ephemeroptera) rep- resenting 20 families were analyzed to obtain a best phylogenetic tree for comparison to previous phylogenetic hypotheses. With respect to relationships among the three major groupings, our mo- lecular data support the hypothesis that and Setisura comprise a monophyletic sister group to the , rather than the hypothesis that Setisura and Furcatergalia comprise a mono- phyletic group stemming from the Pisciforma. Within Pisciforma, acceptable trees show that separates at the base of the Pisciforma clade. The data suggest that Pisciforma is paraphyletic and do not support the grouping of all Southern hemisphere families as a monophyletic group. An evolu- tionary sequence favored by the data suggests a grouping of , Rallidentidae, Nesame- letidae, and and a grouping of Oniscigastridae, Ameletopsidae, and . The data support the monophyly of Setisura (, , Pseudironidae, Oligo- neuriidae, Isonychiidae, and ). Within Setisura, a bootstrap/jackknife test places the families Heptageniidae, Arthropleidae, and Pseudironidae in one clade at 100% frequency. Also supported are hypotheses that Pseudironidae is a sister group to a HeptageniidaeÐArthropleidae group and that a sister relationship exists between the latter two families. Hypotheses that Pseudironidae separated from other Setisura families at an earlier stage and comprises a sister group to a Heptage- niidaeÐOligonuriidae lineage or that Pseudironidae should be moved out of Setisura are not supported.

KEY WORDS Ephemeroptera, 18s rDNA, phylogeny

EPHEMEROPTERA IS ONE OF the most archaic of extant four recent suborders, Furcatergalia, which was some- winged orders (Kukalova´-Peck 1985). The times previously referred to as Rectracheata (McCaf- higher classiÞcation of Ephemeroptera has evolved ferty 1991b), includes , , considerably over the past 200 yr to include progres- all pannote families (), and all tusked bur- sively more higher taxa. However, the use of strict rower families (Scapphodonta); Setisura includes the phylogenetic classiÞcation is relatively recent (Mc- ßat-headed mayßies and their hypothesized minnow- Cafferty 1991b), and cladistic data have been mostly like mayßy ancestral groups, including all forms that morphological in nature. In this study, we used mo- passively Þlter food with their forelegs; Pisciforma lecular data to test the various hypotheses outlined includes all other minnowlike mayßies; and Carapacea below regarding certain aspects of the phylogeny of includes and . Ephemeroptera. Relationship between Carapacea and Other May- Suborder Level Classification of Ephemeroptera. flies. As a distinct monophyletic group supported by At present, four major groupings are generally recog- series of synapomorphies (e.g., the presence of a larval nized in Ephemeroptera under various names. We carapace), the BaetiscidaeÐProsopistomatidae rela- follow a classiÞcation system (Table 1) taken from tionship was suggested by various researchers (Ed- McCafferty (1991b) and its modiÞcations by Wang munds and Traver 1954). Kluge (1998) and McCaf- and McCafferty (1995), McCafferty (1997, 2004), and ferty and Wang (2000) showed the group to be basally McCafferty and Wang (2000). In this classiÞcation of derived with its sister group represented by all other extant mayßies, which share synapomorphic fore- wings with the tornus located between the ends of 1 Human Genome Sequencing Center, Baylor College of Medicine, veins CuA and CuP (CuP and A1 terminating in the Houston, TX 77030. anal margin, rather than the outer margin, of the 2 Department of Biology, Christopher Newport University, wings). McCafferty (1997) Þrst referred to the Baetis- Newport News, VA 23606. 3 Section of Evolution and Ecology, University of California, Davis, cidaeÐProsopistomatidae grouping as the suborder CA 95616. Carapacea. In our study using molecular data, all trees

0013-8746/06/0241Ð0252$04.00/0 ᭧ 2006 Entomological Society of America 242 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

Table 1. McCafferty classification system of extant taxa followed in this study, with families arranged alphabetically within their groupings

Suborder Infraorder Superfamily Family Carapacea Baetiscidae Prosopistomatidae Furcatergalia Lanceolata Leptophlebiidae Palpotarsa Behningiidae Scapphodonta Potamanthoidea Euthyplocioidea Ichthybotidae Pannota Coryphoridae Dicercomyzidae Ephemerythidae Machadorythidae Melanemerellidae Teloganellidae Teloganodidae Vietnamellidae Pisciforma Acanthametropodidae Ameletidae Ameletopsidae Baetidae Dipteromimidae Nesameletidae Oniscigastridae Rallidentidae Siphlaenigmatidae Siphlonuridae Siphluriscidae Setisura Arthropleidae Coloburiscidae Isonychiidae Heptageniidae Pseudironidae that we modiÞed and tested recognize the hypothesis by a series of synapomorphies (McCafferty 1991a, that Baetiscidae, as a representative of Carapacea, is Wang and McCafferty 1995, Kluge 1998). An alterna- the most basal clade. tive hypothesis regarding the relationships among the Phylogenetic Hypotheses Regarding Furcatergalia, three major groupings was given by Kluge et al. (1995) Setisura, and Pisciforma. Among the suborders Fur- and Kluge (1998, 2004), in which the Pisciforma catergalia, Setisura, and Pisciforma, which accommo- group (called Tridentiseta) and a Setisura-Furcater- date the majority of mayßy families, Pisciforma and galia group (called Bidentiseta) were considered sis- Setisura were considered to be derived together, ter clades, although Kluge admitted that Pisciforma constituting a sister clade to the Furcatergalia (Mc- was probably paraphyletic. KlugeÕs grouping of Seti- Cafferty and Edmunds 1979, Landa and Soldan 1985, ´ sura and Furcatergalia together was supported by a McCafferty 1991b, Tomka and Elpers 1991). The presumed initial presence of two dentisetae on larval monophyly of the Furcatergalia and the PisciformaÐ Setisura groups were supported by anatomic synapo- maxillae. Within this grouping, Setisura (or Branchit- morphies, including the presence of the ventral anas- ergaliae) and Furcatergalia (or Furcatergaliae) also tomosis of tracheae in segments other than 8 and 9 of were considered monophyletic groups. Solda´n and the abdomen, and the absence of branches of a dorsal Putz (2000) proposed another hypothesis based on trunk of tracheae formed in the head, respectively karyotype evidence that generally resembled KlugeÕs (Tomka and Elpers 1991). Within the PisciformaÐ hypothesis. It differed from the latter in the arrange- Setisura clade, no clear evidence has supported the ment of lineages within the hypothesized Furcater- monophyly of Pisciforma, and McCafferty and Ed- galiaÐSetisura group by treating Scapphodonta, Lep- munds (1979) indicated it was extremely paraphyletic. tophlebiidae, and Setisura as a monophyletic sister Monophyly of Setisura, however, is clearly supported group to Pannota (Fig. 3). March 2006 SUN ET AL.: PHYLOGENY 243

Phylogeny among Major Groups within Furcater- eages, which were deduced from a systematic study galia. The suborder Furcatergalia is generally thought of mayßy anatomy. Based on internal anatomical data to be monophyletic, with major clades Leptophlebi- and other morphological evidence of Landa and Sol- idae, Scapphodonta, and Pannota containing Ͼ15 fam- da´n (1985), Tomka and Elpers (1991) suggested an ilies. It has been generally recognized that within the evolutionary tree for mayßy families in which most of Furcatergalia, Leptophlebiidae is a basal lineage the clades were supported by presumed synapo- (Kluge 1998), and a sister group to a relatively derived morphies. Within the PisciformaÐSetisura grouping, clade that includes a pair of sister groups, Scapph- Ametropodidae was thought to branch Þrst, with odonta and Pannota (McCafferty and Wang 2000) in other families sharing the nerve ganglion of abdominal addition to a more basal lineage represented by the segment one merging with that of the metathorax. Behningiidae (McCafferty 2004). Scapphodonta and Siphlonuridae (Siphlonuridae and Dipteromimidae Pannota are both monophyletic and are supported by in Table 1) and Rallidentidae (Rallidentidae, Nesa- a series of morphological synapomorphies (McCaf- meletidae, and Ameletidae in Table 1) were thought ferty and Wang 2000, McCafferty 2004). In our study, to be sister groups that branched next, with other we followed this arrangement of branches within Fur- families lacking a nerve ganglion in abdominal seg- catergalia in our construction of hypothesized phylo- ment 8. Acanthametropodidae and Ameletopsidae genetic trees. were thought to branch next from the remainder that Relationships Regarding the Pisciforma–Setisura included Baetidae, Metretopodidae, and Setisura, Grouping. When Ulmer (1920) Þrst introduced sub- which supposedly shared paired intercalaries in the order concepts to Ephemeroptera, all minnowlike forewings parallel with CuA and CuP. Baetidae was mayßies (except Baetidae) and ßatheaded mayßies treated as a sister group to a MetretopodidaeÐHep- were placed in the suborder Heptagenioidea. Ed- tagenioidea grouping, that shared two-segmented la- munds and Traver (1954) changed UlmerÕs suborders bial palpi. Metretopodidae was considered to be a into superfamilies, while adding Baetidae and Oligo- sister group to Heptagenioidea. However, the conclu- neuriidae to Heptagenioidea. Edmunds et al. (1963) sions of Landa and Solda´n (1985) and Tomka and considered Isonychiinae and Coloburiscinae as sub- Elpers (1991) have been suspect because the char- families of Siphlonuridae (in Heptagenioidea), and acters used in their analyses have a propensity for Riek (1973) placed them in Oligoneuriidae, which was convergence and are inconsistent within the taxa later followed by McCafferty and Edmunds (1979). they were said to represent, as have been pointed Demoulin (1958) proposed Siphlonuroidea for Siph- out, for example, by Kluge et al. (1995) and McCaf- lonuridae and Baetidae to recognize their hypothe- ferty (2004). sized close relationship, and he also separated them Kluge et al. (1995) recognized two superfamilies from the ßatheaded mayßies (Heptageniidae and in Pisciforma: Baetoidea (including Baetidae and some Oligoneuriidae), which he placed in Heptage- Siphlaenigmatidae) and Siphlonuroidea (including nioidea and Oligoneurioidea. McCafferty and Ed- the rest of the Pisciforma families). These authors munds (1979) continued to include minnowlike and also divided Siphlonuroidea into a Northern hemi- ßatheaded mayßies together in one superfamily but sphere group of families that included Siphlon- changed the name from Heptagenioidea to Baetoidea; uridae, Dipteromimidae, Ameletidae, Metretopodi- however, Landa and Solda´n (1985) used Baetoidea for dae, Acanthametropodidae, and Ametropodidae, and only minnowlike mayßy families and Heptagenioidea a Southern hemisphere group of families that included for Heptageniidae and Oligoneuriidae. McCafferty Oniscigastridae, Nesameletidae, Rallidentidae, and (1990) Þrst considered the Coloburiscidae, Isonychi- Ameletopsidae and suggested that the Southern hemi- idae, Oligoneuriidae, and Heptageniidae to constitute sphere group was possibly monophyletic. the Heptagenioidea, and later further elaborated cla- Relationships among Families of Setisura. Various distic evidence for the monophyly of his grouping researchers have believed that Setisura represented with numerous synapomorphies (McCafferty 1991a). an evolutionary branch derived from Pisciforma McCafferty (1991b) proposed suborders Setisura and mayßies that became adapted to ßowing water habi- Pisciforma for Heptagenioidea and Baetoidea, respec- tats by way of passive Þlter feeding (Sinitshenkova tively, in which a series of mayßy subfamilies previ- 1984, Edmunds and McCafferty 1988, McCafferty ously placed under Siphlonuridae or Oligoneuriidae 1991a). The determination of which families repre- were treated as families. Later, Wang and McCafferty sent the intermediate evolutionary links is most im- (1995) moved Pseudironidae from Siphlonuroidea to portant for establishing a phylogenetic classiÞcation. Heptagenioidea. However, Kluge (2004) suggested Within Setisura, McCafferty (1990, 1991a) hypothe- that Pseudironidae should be moved out of Setisura, sized an evolutionary sequence from the primitive being instead a pisciform group. forms, including Coloburiscidae, Isonychiidae, and Relationships among Families of Pisciforma. The Oligoneuriidae, to the more derived Heptageniidae evolution of the families of Pisciforma still remains based on morphological evidence from fossil and ex- largely unclariÞed. Edmunds (1975), Landa (1973), tant mayßies. Wang and McCafferty (1995) also hy- and Riek (1973) all presented various hypotheses pothesized that Pseudironidae was relatively highly of family evolution that included pisciform taxa derived within Setisura, representing a sister group based on phenetic interpretations. Landa and Solda´n to the HeptageniidaeÐArthropleidae group, all of (1985) discussed evolutionary trends of many lin- which represented the most apotypic clade in the 244 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

Table 2. Mayfly taxa sequenced for this study

Family Species Collection locale Acanthametropodidae eximia Edmunds Canada: Saskatchewan Ameletidae celer McDunnough USA: Montana Ameletopsidae Chiloporter sp. Chile: Region IX Ametropodidae Ametropus neavei McDunnough Canada: Saskatchewan Arthropleidae bipunctata (McDunnough) USA: Michigan Baetidae Baetis bicaudatus Dodds USA: Montana Callibaetis ferrugineus (Walsh) USA: Montana Baetiscidae lacustris McDunnough USA: Montana Caenidae Caenis youngi Roemhild USA: Montana Coloburiscidae Coloburiscoides giganteus Tillyard Australia: Australian Capital Territory Ephemeridae Ephemera simulans Walker USA: Indiana Heptageniidae grandis (McDunnough) USA: Montana diabasia Burks USA: Nebraska Isonychiidae rufa McDunnough USA: Indiana Leptophlebiidae Paraleptophlebia memorialis (Eaton) USA: Montana Metretopodidae Metretopus borealis (Eaton) Canada: Northwest Territory Nesameletidae Ameletoides lacusalbinae Tillyard Australia: New South Wales Oligoneuriidae Lachlania talea Allen & Cohen Honduras: Olancho Oniscigastridae Tasmanophlebia sp. Australia: New South Wales Polymitarcyidae Tortopus primus (McDunnough) USA: Indiana Pseudironidae centralis McDunnough USA: Wisconsin Siphlonuridae columbianus McDunnough USA: Montana suborder. Alternative evolutionary relationships of later use. Total genomic DNA was extracted using Pseudironidae and other Setisura families were sug- InstaGene Matrix (Bio-Rad, Hercules, CA) according gested by Tomka and Elpers (1991), in which Pseud- to manufacturerÕs recommendations. The mixtures ironidae was thought to separate from other Setisura were incubated overnight at 56ЊC and then were at an earlier stage and was a sister taxon to a hypoth- boiled and centrifuged as directed by the manufac- esized HeptageniidaeÐOligoneuriidae lineage. Also, as turer. The supernatant was stored at 4ЊC for use as mentioned above, Kluge (2004) did not consider templates for polymerase chain reaction (PCR). Prim- Pseudironidae to be a member of Setisura. ers (Ferris et al. 2004) for the Ϸ620-bp fragment of the 18s rDNA were (forward) 5Ј-AGGGCAAGTCTGGT- GCCAGC-3Ј and (reverse) 5Ј-TTTCAGCTTTGCA- Materials and Methods ACCATAC-3Ј. The ampliÞed fragment was cloned Specimens. We sequenced partial 18s rDNA from 22 into the pGEM-T vector (Promega, Madison, WI) mayßy species belonging to 20 families as shown in and transformed into Escherichia coli strain JM109 Table 2. Specimens were collected between 1982 and (Promega). Colony cultures containing inserts of the 2002, preserved in 70% ethanol, and deposited in the expected size, as determined by PCR, were used to Purdue Entomological Research Collection (PERC). make Wizard Plus plasmid preparations (Promega). The homologous sequence of one species of Ralliden- Automatic sequencing of the plasmid preparations tidae, Rallidens mcfarlanei Penniket (GenBank acces- was done at the Purdue Genomics Facility. Both sion no. AY338696), was incorporated into our anal- strands of DNA from two to four clones were se- yses. Therefore, all Pisciforma and Setisura families quenced for each insect isolate. were included in this study except Siphlaenigmatidae, Methods for Phylogenetic Analysis, Tree Building, Siphluriscidae, and Dipteromimidae. These three fam- and Testing. Sequences were aligned using Clustal X ilies are thought to have close relationships with Baeti- (Thompson et al. 1997) with default parameters (gap dae, Acanthametropodidae (or possibly Nesameleti- open ϭ 15, gap extend ϭ 6.66). Phylogenetic analysis dae), and Siphlonuridae, respectively, based on was performed with PAUP* 4.0b10 (Swofford 2002), morphological evidence (Edmunds and Koss 1972, under Parsimony criteria, by using the heuristic search Edmunds 1973, Meyer et al. 2003, Zhou and Peters with Lepisma sp. assigned as an outgroup. We used 2003). We used one species of Thysanura (Lepisma sp., MacClade 4.01 (Maddison and Maddison 2000) to GenBank accession no. AF005458) to serve as the create trees reßective of hypotheses for the phylogeny non-Ephemeroptera outgroup in our molecular anal- of Ephemeroptera based on morphological data dis- yses. cussed above. Hypotheses were compared individu- Methods for Obtaining and Manipulating DNA. We ally to the best tree based on molecular data by using dissected and rinsed in Tris-EDTA buffer (TE, pH 7.5) the ShimodairaÐHasegawa (S/H) test as implemented pieces of muscle from the thorax or abdomen of a in PAUP. In this test, when P Ն 0.05, trees were single specimen of each mayßy species. We used a considered not signiÞcantly different from the original Kontes grinder (Kontes Glass, Vineland, NJ) to grind best tree, and the hypotheses represented by these the tissue in 25 ␮l of Molecular Grinding Resin (Geno- trees were not rejected. When P Ͻ 0.05, the trees were tech, St. Louis, MO) in a 1.5-ml microcentrifuge tube, rejected. In addition, we manipulated these trees in which was immediately used or frozen at Ϫ20ЊC for PAUP and MacClade to obtain additional shorter trees March 2006 SUN ET AL.: MAYFLY PHYLOGENY 245

Fig. 1. Tree 1: an acceptable tree that supports the hypothesis of relationships among the three major groupings of mayßies, in which Pisciforma and Setisura comprise a monophyletic sister group to Furcatergalia. with better S/H scores if possible for the hypothesis PAUP with Thysanura as the outgroup. The family being tested. For example, we frequently searched Coloburiscidae, hypothesized as the most primitive partial trees containing only some of the groups to Setisura mayßy form (McCafferty 1991a), served as a obtain optimal arrangements for them. positional indicator of the origin of the Setisura sub- order within Pisciforma. Taxa of Pisciforma were ar- ranged in MacClade according to this partial tree. The Results and Discussion remaining lineages of Setisura were added on the Relationships among Setisura, Pisciforma, and Fur- Coloburiscidae branch and were placed in the search catergalia. The Þrst hypothesis with respect to rela- for the shortest total length function in MacClade. tionships among the three major groupings is that When the Þrst hypothesis and other assumptions de- Pisciforma and Setisura comprise a monophyletic sis- scribed above were followed, the tree with the highest ter group to Furcatergalia. In the test of this hypoth- possible score is shown in Fig. 1. This tree was not esis, the best tree was modiÞed in the following ways: signiÞcantly different from the best molecular tree Baetiscidae was moved to the base of the clade that based on the S/H test score. Therefore, our data sup- represented all extant Ephemeroptera. Above the ported the Þrst hypothesis regarding the relationships Baetiscidae branch, all Furcatergalia taxa were placed among the three major groupings of mayßies, repre- at the base of the three major lineages to form one sented by the framework of McCafferty (1991b). clade, which was a sister group to a hypothesized The alternative hypothesis is that the suborders PisciformaÐSetisura clade. Within Furcatergalia, ar- Setisura and Furcatergalia comprise a monophyletic rangements of families followed hypotheses that sister group. However, the ancestral nature of this were described above: Letophlebiidae separated at an group has not been clearly inferred in this hypothesis. earlier stage at the base of the lineage; the sister Therefore, we assumed that there are two possibilities: groups, Pannota (represented by Caenidae) and Scap- the hypothesized ancestor could be either a Setisura- phodonta (represented by Ephemeridae and Poly- like form, or a Furcatergalia-like form. To obtain the mitarcyidae) were placed at the top. In the Pisci- best tree score, the above-mentioned possibilities formaÐSetisura clade, Setisura was considered to be were examined separately in two trees. In the Þrst tree, monophyletic. To obtain the tree with the highest the ancestor of the hypothesized grouping of Setisura possible score, a partial tree that included only data and Furcatergalia was represented by Coloburiscidae, of the Pisciforma taxa plus Coloburiscidae was in- which was considered to be the most primitive form ferred using the Heuristic Search under Parsimony in in Setisura, whereas in the second tree it was repre- 246 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

Fig. 2. Trees 2 and 3: best trees following the second hypothesis regarding relationships among the three major groupings of mayßies, in which Furcatergalia and Setisura comprise a monophyletic group stemming from Pisciforma. Tree 2: a tree built on the assumption that the ancestor of the hypothesized FurcatergaliaÐSetisura lineage is a Setisura-like form. Tree 3: a tree built on the assumption that the ancestor of the hypothesized FurcatergaliaÐSetisura lineage is a Furcatergalia-like form. Both trees were rejected by the S/H test. sented by Leptophlebiidae, which was considered to further searches for the shortest total length in Mac- be the most primitive form in Furcatergalia. After Clade by using the taxa of the FurtcatergaliaÐSetisura placing Baetiscidae at the base of the entire Ephemer- group of trees 2 and 3. However, all trees generated optera clade, we inferred using the Heuristic Search using this method were rejected by the S/H tests. under Parsimony a partial tree that included only data Therefore, our molecular data did not support the of Pisciforma taxa plus Coloburiscidae, with Thysa- second hypothesis, represented by the framework of nura as the outgroup. Next, taxa of Pisciforma were Kluge (1998, 2004). arranged above Baetiscidae in MacClade as indicated We also examined the hypothesis given by Solda´n by the partial tree. The rest of the lineages of Setisura and Putz (2000) based on karyotype evidence. and Furcatergalia were then added on the Colo- Trees 2 and 3 (Fig. 2) were modiÞed by moving Eph- buriscidae branch. Taxa of these two suborders were emeridae, Polymitarcyidae, and Leptophlebiidae from placed in two respective clades as sister groups. In Furcatergalia to Setisura as suggested by Solda´n and each of the clades, families and genera were arranged Putz (2000). The S/H test showed these trees to be in the same way as for tree 1 (Fig. 1). The tree created signiÞcantly different from the best tree. We also is shown in Fig. 2 (tree 2). Second, a partial tree performed the Heuristic Search under Parsimony cri- including only data of all Pisciforma taxa plus Lepto- phlebiidae was inferred using the Heuristic Search teria and obtained a best tree based on the molecular under Parsimony with Thysanura as the outgroup. data for only the families studied by Solda´n and Putz Another complete tree was constructed based on (2000) (except Potamanthidae and Ephemerellidae) data from this partial tree in the same way as that with Thysanura as the outgroup. We used MacClade described above, except that the hypothesized Fur- to create another tree that included the same may- catergaliaÐSetisura clade was added on the Leptophle- ßy families (Baetidae, Siphlonuridae, Heptageniidae, biidae branch instead of the Coloburiscidae branch Oligoneuriidae, Leptophlebiidae, Ephemeridae, and (Fig. 2, tree 3). Trees two and three were compared Caenidae) following the arrangement of Solda´n and with the best tree based on molecular data (using the Putz (2000) (Fig. 3). The S/H test indicated a signif- S/H test as implemented in PAUP), and both trees icant difference between this tree and the best tree were signiÞcantly different from the best tree. As a based on molecular data. Therefore, the framework further test of the possibility that Furcatergalia and suggested by Solda´n and Putz (2000) was not sup- Setisura comprise a monophyletic group, we made ported by our molecular data. March 2006 SUN ET AL.: MAYFLY PHYLOGENY 247

Fig. 3. Tree 4: a tree created including limited families following the hypothesis given by Solda´n and Putz (2000), in which Setisura, tusked burrowers, and Leptophlebiidae were considered to form a monophyletic group. This tree was rejected by the S/H test when compared with the best tree based on our molecular data of the same families.

According to our results, Setisura evolved indepen- We then examined the possibility that Siphlonuri- dently from a siphlonurid-like ancestor within Pisci- oidea included a monophyletic Southern hemisphere forma and did not share a direct common ancestor group comprised of Oniscigastridae, Nesameletidae, with Furcatergalia, which originated independently Rallidentidae, and Ameletopsidae as suggested by within Ephemeroptera. Because the PisciformaÐSeti- Kluge et al. (1995). Tree 1 was modiÞed so that the sura (PisciformaÐHeptagenioidea) clade is monophy- four families comprised one clade. We used MacClade letic, subordinal status could be assigned to this entire to Þnd an arrangement with the shortest total length group. As such, Setisura would be subsumed under within this clade and tentatively added it at the orig- Pisciforma, as per McCafferty (1997). inal position of each of these families to explore all Status of Pisciforma. Additional trees were gen- possibilities (e.g., Fig. 4, tree 7). However, all such erated in MacClade to examine the status of Pisci- trees were rejected by the S/H test, and our molecular forma, sensu stricto. These trees were modiÞed from data suggest that all Southern hemisphere Siphlonu- tree 1 (Fig. 1) by adjusting the positions of Pisci- roidea families are not likely to be derived from a forma families, which included Acanthametropo- single common ancestor. didae, Ameletidae, Ameletopsidae, Ametropodidae, Following the assumption that Pisciforma is para- Baetidae, Metretopodidae, Nesameletidae, Onis- phyletic, a series of trees was constructed to test the cigastridae, Rallidentidae, Siphlaenigmatidae, and phylogeny of this group as hypothesized by Tomka Siphlonuridae. and Elpers (1991) as shown in tree 8 (Fig. 4). Because We Þrst explored the possibility that Pisciforma is the family Oniscigastridae was not included in the monophyletic. A partial tree was obtained by includ- Tomka and Elpers study, we obtained another tree ing data of all Pisciforma taxa (with Thysanura as that included molecular data for all our taxa except outgroup) by using the Heuristic Search under Par- Oniscigastridae by using the Heuristic Search under simony. We then selected the partial tree with the best Parsimony. This best molecular tree was compared score and modiÞed tree 1 (Fig. 1) by placing all taxa with a series of trees created to examine the various of Pisciforma in one clade that followed the arrange- aspects of the hypothesis. As suggested by Tomka and ment indicated by the best partial tree. Baetidae is at Elpers (1991), we Þrst treated Metretopodidae as a the base of the Pisciforma clade as in tree 1 (Fig. 1), sister group to Setisura. A partial tree was obtained by whereas all other families (Siphlonuridae-like forms) including data of all Pisciforma taxa (with Thysanura form a sister clade (Fig. 4, tree 5). The S/H test as outgroup) based on molecular data using the Heu- indicated that the new tree was signiÞcantly different ristic Search under Parsimony in PAUP. Tree 1 (Fig. 1) from the best tree based on molecular data. We also was modiÞed in MacClade, with the families of Pisci- explored the possibility that all Siphlonuridae-like forma arranged as indicated by the partial tree ob- families comprised a monophyletic group (Siphlonu- tained previously, whereas Setisura was added on the roidea) when a paraphyletic Pisciforma was divided Metretopodidae branch as a sister group to the latter. into two superfamilies: Baetoidea (represented in our The S/H test showed the resulting tree to be signiÞ- test by Baetidae) and Siphlonuroidea (represented by cantly different from the best tree. all other Pisciforma families). Another tree was con- Next, we examined the possibility that Setisura, structed based on tree 5, in which all Pisciforma (ex- Baetidae, and Metretopodidae comprise a monophy- cluding Baetidae) were placed in one clade as a sister letic group. Another partial tree that included all group to Setisura. This tree (Fig. 4, tree 6) also was Pisciforma taxa except Metretopodidae was obtained rejected by the S/H test, and therefore our molecular based on molecular data using PAUP. Arrangements data suggest that Pisciforma is paraphyletic. If it is of these families in tree 1 (Fig. 1) were modiÞed as divided into two superfamilies, Siphlonuroidea in it- indicated in this new partial tree, and Metretopodidae self is also a paraphyletic group. and Setisura were added on the Baetidae branch. The 248 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

Fig. 4. Trees 5Ð8. Tree 5: the best tree following the assumption that Pisciforma is a monophyletic group. Tree 6: the best tree following the assumption that Pisciforma could be divided into two separate monophyletic groups, Baetoidea and Siphlonuroidea. Tree 7: an example of trees following the assumption that Pisciforma included a monophyletic Southern hemisphere family group. Tree 8: a hypothesized tree in which the arrangement of Pisciforma families follows the hypothesis given by Tomka and Elpers (1991), where Nesameletidae and Ameletidae were genera under Rallidentidae. All of these trees were rejected by the S/H test. clade comprised by these three groups also was placed function in MacClade. The second partial tree in- in the search-above-branch function to obtain the cluded the same taxa as in the Þrst partial tree except highest possible score. This tree also was rejected by that Acanthametropodidae was replaced by Amele- the S/H test. topsidae. Accordingly, the group of interest was added We then constructed another tree to examine the on the Ameletopsidae branch in MacClade. Both these possibility that Setisura, Baetidae, Metretopodidae, trees were shown by the S/H test to be signiÞcantly Acanthametropodidae, and Ameletopsidae comprise different from the best tree. a monophyletic group as suggested by Tomka and We examined whether Ametropodidae could be the Elpers (1991). Because they suggested that Acan- most primitive form of the PisciformaÐSetisura lin- thametropodidae and Ameletopsidae were a sister eage, as suggested by Tomka and Elpers (1991), by group and formed a most primitive clade in this group, placing the family at the base of the clade that in- we placed each of two families at the base of this cluded all PisciformaÐSetisura families in tree 1 hypothesized group. The Þrst partial tree included (Fig. 1), and using the search for the shortest total only Acanthametropodidae plus the rest of the Pisci- length function in MacClade on all taxa above the forma taxa. The arrangements of these families were Ametropodidae branch. The tree generated was re- modiÞed in tree 1 following this partial tree, and then jected according to the S/H test. Therefore, we con- a group of taxa including Ameletopsidae, Baetidae, clude that several characteristics treated as synapo- Metretopodidae, and Setisura was added to the Acan- morphies by Tomka and Elpers (1991) in their thametropodidae branch as a monophyletic group hypothesized phylogeny of the SetisuraÐPisciforma and placed in the search for the shortest total length group are the result of homoplasy. These character- March 2006 SUN ET AL.: MAYFLY PHYLOGENY 249

Fig. 5. Tree 9: an acceptable tree based on tree 1 following one hypothesis given by Landa and Solda´n (1985), in which Ameletopsidae and Oniscigastridae are sister groups, and two hypotheses given by Tomka and Elpers (1991): 1) Present Siphlonuridae, Rallidentidae, Nesameletidae, and Ameletidae (of Table 1) comprise a monophyletic group; and 2) Amele- topsidae and Acanthametropodidae are sister groups (when Oniscigastridae was absent). istics include the presence of two-segmented labial (with Thysanura as the outgroup), placement of palps, the presence of paired intercalaries parallel with Baetidae at the base of this clade was supported by a CuA and CuP on forewings, the absence of a nerve 100% bootstrap score. This suggested that Baetidae ganglion in abdominal segment 8, and the merging of (the superfamily Baetoidea, including Baetidae and the nerve ganglion of abdominal segment 1 and that of possibly Siphlaenigmatidae) separated at a very early the metathorax. stage from other lineages of Pisciforma, leaving a We tested two additional aspects of the Tomka paraphyletic superfamily Siphlonuroidea. The combi- and Elpers (1991) hypothesis. In their study, Siph- nation of the rest of the Pisciforma families plus all lonuridae and Rallidentidae (Rallidentidae, Nesame- the Setisura families into one monophyletic group had letidae, and Ameletidae in Table 1) were thought to a bootstrap support of 52%, as did the arrangement of comprise a monophyletic group, although no synapo- Rallidentidae and Ameletidae as a sister group in morphy was provided. They also suggested that Acan- Pisciforma. thametropodidae and Ameletopsidae are sister groups Evolution of Setisura. At the outset of our study, the supported by anatomic evidence. The trees that we suborder Setisura was considered by many to be a created that reßected these hypotheses could not be monophyletic group that included the families Hep- rejected based on the S/H tests. We further added Oniscigastridae back to this tree. Because the family tageniidae, Arthropleidae, Pseudironidae, Oligoneuri- was not included in the Tomka and Elpers (1991) idae, Coloburiscidae, and Isonychiidae (Wang and study, we followed the hypothesis given by Landa and McCafferty 1995, Kluge 1998). Our data strongly sup- Solda´n (1985), in which Ameletopsidae and Onis- ported the hypothesis that Heptageniidae, Arthro- cigastridae were considered to be sister groups. Thus, pleidae, and Pseudironidae form a monophyletic in our tree Acanthametropodidae was placed at the group. These three families (represented by four se- base of a clade comprised by the two families (Fig. 5, quences) were exclusively placed in one clade at 100% tree 9). The S/H test indicated that this new tree was bootstrap frequency on the best tree based on mo- not signiÞcantly different from the best tree. There- lecular data that included all our mayßy sequences. As fore, these particular aspects of the Tomka and Elpers discussed above, our S/H test results indicated that it (1991) hypothesis are acceptable. was acceptable to include the families Oligoneuriidae, On the best phylogenetic tree based on molecular Coloburiscidae, and Isonychiidae in Setisura, thus sup- data of the families of the SetisuraÐPisciforma group porting the monophyly of this suborder that was based 250 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

Fig. 6. Trees 10Ð12. Tree 10: an alternative hypothesis regarding the evolution of Setisura lineages, in which Coloburisci- dae, Isonychiidae, and Oligoneuriidae were placed in one clade. Tree 11: an alternative evolutionary hypothesis of Setisura, in which Pseudironidae is a sister group to a hypothesized OligonuriidaeÐHeptageniidaeÐArthropleidae lineage. Tree 12: a parsimonious arrangement of Pisciforma and Setisura lineages based on the hypothesis that Pseudironidae is a part of Pisciforma, not Setisura. All of these trees were rejected by the S/H test. on morphological evidence, as per the framework of shortest total length function in MacClade. The S/H Wang and McCafferty (1995). test indicated that this tree was signiÞcantly different The sequence of evolution of the Setisura lineages from the best tree and therefore not acceptable based shown in tree 1 (Fig. 1) and tree 9 (Fig. 5) was a result on our molecular data. Thus, our molecular evidence of the search for the shortest total length function in suggested that Coloburiscidae, Isonychiidae, and Oli- MacClade, and favored the hypothesis given by Mc- goneuriidae separated from the rest of Setisura at Cafferty (1991a) based on morphological evidence, different times and do not comprise a monophyletic regarding the origins of Setisura and relationships group, and the explanation for the two incongruent among Coloburiscidae, Isonychiidae, Oligoneuriidae, characters, as given by McCafferty (1991a), is highly and Heptageniidae. McCafferty (1991a) also had ob- probable, which also was supported by data from ecol- served that two of the 36 apomorphies he studied ogy and behavior (McCafferty 1991a). occurred exclusively in the families Coloburiscidae, Our data favored the hypothesis by Wang and Mc- Isonychiidae, and Oligoneuriidae, all possessing apo- Cafferty (1995) regarding the relationship between morphic maxillary gills and Þltering setae on the fore- Pseudironidae and the HeptageniidaeÐArthropleidae legs. McCafferty (1991a) hypothesized that these grouping (Figs. 1 and 4). To examine the accept- were gained synapomorphies at the base of Setisura ability of alternative hypotheses regarding sequence that were subsequently lost in the apotypic family of evolution of these families, we created additional Heptageniidae. Nonetheless, an alternative arrange- trees. Tree 1 (Fig. 1) was modiÞed by moving Pseud- ment of lineages in a phylogeny might be to place ironidae to the base of the clade comprised of Hep- these three families exclusively in a single clade to tageniidae, Arthropleidae, and Oligoneuriidae (Fig. 6, represent a primitive lineage of Setisura. We con- tree 11) as suggested by Tomka and Elpers (1991). To structed a tree modiÞed from our tree 1 (Fig. 1), in examine the hypothesis given by Kluge (2004), who which such a clade was recognized (Fig. 6, tree 10). suggested that Pseudironidae should be moved out of The rest of the Setisura families, Pseudironidae, Ar- Setisura, a partial tree that included only Pseud- thropleidae, and Heptageniidae, were placed in an- ironidae, Coloburiscidae, and all the Pisciforma fam- other clade as a sister group to the former. This group- ilies was inferred using the Heuristic Search under ing of the three lineages was supported with the 100% Parsimony in PAUP, with Thysanura as the outgroup. bootstrap frequency and also was supported by mor- The complete tree was constructed by modifying tree phological synapomorphies noted by McCafferty 1 (Fig. 1), in which the Pseudironidae and Pisciforma (1991a) and Wang and McCafferty (1995). Each of taxa were arranged as indicated by the new partial these clades was then placed in the search for the tree; the Setisura taxa (except Pseudironidae) were March 2006 SUN ET AL.: MAYFLY PHYLOGENY 251

Fig. 7. Tree 13: a tree modiÞed from tree 9 with Arthropleidae a sister group to Heptageniidae. This tree was accepted by the S/H test. added on the Coloburiscidae branch and placed in very long time spans. These variations can lead to another search for the shortest total length in Mac- errors in estimates of phylogenetic relationships in the Clade (Fig. 6, tree 12). Both tree 11 and tree 12 (Fig. 6) same way that long time spans may obscure pathways were rejected by the S/H tests. Therefore, our mo- of changes in morphological characters. To further lecular data support the hypothesis that Heptageni- clarify the phylogeny of Ephemeroptera, more re- idae, Arthropleidae, and Pseudironidae comprise a search with both morphological and molecular data clade separate from Oligoneuriidae within the Hep- sets is needed. tagenioidea. Wang and McCafferty (1995) discussed the sister group relationship between Heptageniidae and Ar- Acknowledgments thropleidae, which was supported by morphological We thank J. Webb (Purdue University, West Lafayette, synapomorphies. We tested this hypothesis by mod- IN) for collecting part of the mayßy specimens. We also ifying tree 9 (Fig. 5) to make another tree in which thank J. S. Yaninek (Purdue University) and the Department Heptageniidae genera Heptagenia and Epeorus were of Entomology for supporting this project, which was initi- placed in one clade to represent the family; Arthro- ated during a Molecular Systematics course in which we pleidae was treated as a sister group to the Heptage- participated. niidae clade (Fig. 7, tree 13). The S/H test indicated no signiÞcant difference from the best tree. Therefore, our data support a sister group relationship between References Cited Heptageniidae and Arthropleidae. Demoulin, G. 1958. Nouveau schema de classiÞcation des Our molecular data support tree 13 (Fig. 7) regard- Archodonates et des Ephemeropteres. Bull. Inst. R. Sci. ing the phylogeny of Ephemeroptera. Other trees are Nat. Belg. 39: 1Ð18. probably also acceptable, especially with respect to Edmunds, G. F., Jr. 1973. Some critical problems of family evolution among Pisciforma families. Uncertainty in relationships in the Ephemeroptera, pp. 145Ð154. In W. L. the reconstruction of a phylogenetic tree for mayßies Peters and J. G. Peters [eds.], Proceedings of the First International Conference on Ephemeroptera. E. J. Brill, probably stems from their long evolutionary history, Leiden, The Netherlands. being an ancient group of winged . Despite the Edmunds, G. F., Jr. 1975. Phylogenetic biogeography of often-demonstrated highly conserved nature of 18S mayßies. Ann. Mo. Bot. Gard. 62: 251Ð263. ribosomal DNA (Ferris et al. 2004), variations in rates Edmunds, G. F., Jr., and R. W. Koss. 1972. A review of the of molecular change in the rRNA gene may occur over Acanthametropodinae with a description of a new 252 ANNALS OF THE ENTOMOLOGICAL SOCIETY OF AMERICA Vol. 99, no. 2

(Ephemeroptera: Siphlonuridae). Pan-Pac. Entomol. 48: non-holometabolous orders. Entomological Information 136Ð144. Services, Rockville, MD. Edmunds, G. F., Jr., and W. P. McCafferty. 1988. The mayßy McCafferty, W. P. 2004. Higher classiÞcation of the subimago. Annu. Rev. Entomol. 33: 509Ð529. burrowing mayßies (Ephemeroptera: Scapphodonta). Edmunds, G. F., Jr., and J. R. Traver. 1954. An outline of a Entomol. News 115: 84Ð92. reclassiÞcation of the Ephemeroptera. Proc. Entomol. McCafferty W. P., and G. F. Edmunds, Jr. 1979. The higher Soc. Wash. 56: 236Ð240. classiÞcation of the Ephemeroptera and its evolutionary Edmunds, G. F., Jr., R. K. Allen, and W. L. Peters. 1963. An basis. Ann. Entomol. Soc. Am. 72: 5Ð12. annotated key to the nymphs of the families and subfam- McCafferty, W. P., and T.-Q. Wang. 2000. Phylogenetic sys- ilies of mayßies (Ephemeroptera). Univ. Utah Biol. Ser. tematics of the major lineages of pannote mayßies 13: 1Ð49. (Ephemeroptera: Pannota). Trans. Am. Entomol. Soc. Ferris, V. R., A. Sabo, J. G. Baldwin, M. Mundo-Ocampo, 126: 9Ð101. R. N. Inserraa, and S. Sharma. 2004. Phylogenetic rela- Meyer, M. D., W. P. McCafferty, and E. L. Silldorff. 2003. tionships among selected Heteroderoidea based on 18S Relationships of Edmundsius Day (Ephemeroptera: and ITS ribosomal DNA. J. Nematol. 36: 202Ð206. Siphlonuridae), with notes on early instar larvae and Kluge, N. J. 1998. Phylogeny and higher classiÞcation of eggs. Pan-Pac. Entomol. 79: 112Ð118. Ephemeroptera. Zoosyst. Ross. 7: 255Ð269. Riek, E. F. 1973. ClassiÞcation of the Ephemeroptera, Kluge, N. J. 2004. The phylogenetic system of Ephemerop- pp. 160Ð178. In W. L. Peters and J. G. Peters [eds.], tera. Kluwer Academic Publishers, Dordrecht, The Neth- Proceedings of the First International Conference on erlands. Ephemeroptera. E. J. Brill, Leiden, The Netherlands. Kluge, N. J., D. Studemann, P. Landolt, and T. Gonser. 1995. Sinitshenkova, N. D. 1984. The Mesozoic mayßies A reclassiÞcation of Siphlonuroidea (Ephemeroptera). (Ephemeroptera) with special reference to their ecology, Bull. Soc. Entomol. Suisse 68: 103Ð132. pp. 61Ð66. In V. Landa, T. Solda´n, and M. Toner [eds.], Proceedings of the Fourth International Conference on Kukalova´-Peck, J. 1985. Ephemeroid wing venation based Ephemeroptera, Czechoslovak Academy of Sciences, upon new gigantic mayßies and basic mor- Prague, Czech Republic. phology, phylogeny, and metamorphosis of pterygote in- Soldan, T., and M. Putz. 2000. Karyotypes of some central sects (Insecta, Ephemerida). Can. J. Zool. 63: 933Ð955. ´ European mayßies (Ephemeroptera) and their contribu- Landa, V. 1973. A contribution to the evolution of the order tion to phylogeny of the order. Acta Soc. Zool. Bohem. 64: Ephemeroptera based on comparative anatomy, pp. 155Ð 437Ð445. 159. In W. L. Peters and J. G. Peters [eds.], Proceedings Swofford, D. L. 2002. PAUP*: phylogenetic analysis of the First International Conference on Ephemeroptera. using parsimony (* and other methods), version 4.0b10. E. J. Brill, Leiden, The Netherlands. Sinauer, Sunderland, MA. Landa, V., and T. Solda´n. 1985. Phylogeny and higher clas- Thompson, J. D., T. J. Gibons, F. Plewnik, F. J. Mougin, and siÞcation of the order Ephemeroptera: a discussion from D. G. Higgins. 1997. The Clustal X Windows interface: the comparative anatomical point of view. Academia, ßexible strategies for multiple sequence alignment aided Cesk. Akad. Ved., Praha, Czech Republic. by qualitative analysis tools. Nucleic Acids Res. 25: 4876Ð Maddison, D. R., and W. P. Maddison. 2000. MacClade 4: 4882. analysis of phylogeny and character evolution, version Tomka, I., and C. Elpers. 1991. Problems in the phylogeny 4.0. Sinauer, Sunderland, MA. of the Ephemeroptera, pp. 115Ð134. In J. Alba-Tercedor McCafferty, W. P. 1990. Chapter 2. Ephemeroptera, pp. 20Ð and A. Sanchez-Ortega [eds.], Overview and strategies of 50. In D. A. Grimaldi [ed.], Insects from the Santana Ephemeroptera and Plecoptera. The Sandhill Crane Formation, Lower , of Brazil. Bull. Am. Mus. Press, Inc., Gainesville, FL. Nat. Hist. 195. Ulmer, G. 1920. Neue Ephemeropteren. Arch. Naturges. McCafferty, W. P. 1991a. The cladistics, classiÞcation, and (1919) 85 (Abt. A, Heft 11): 1Ð80. evolution of the Heptagenioidea, pp. 87Ð101. In J. Alba- Wang, T.-Q., and W. P. McCafferty. 1995. Relationships of Tercedor and A. Sanchez-Ortega [eds.], Overview and the Arthropleidae, Heptageniidae, and Pseudironidae strategies of Ephemeroptera and Plecoptera. The San- (Ephemeroptera: Heptagenioidea). Entomol. News 106: dhill Crane Press, Inc., Gainesville, FL. 251Ð256. McCafferty, W. P. 1991b. Toward a phylogenetic classiÞca- Zhou, C.-F., and J. G. Peters. 2003. The nymph of Siphlu- tion of the Ephemeroptera (Insecta): a commentary on riscus chinensis and additional imaginal description: a liv- systematics. Ann. Entomol. Soc. Am. 84: 343Ð360. ing mayßy with origins (Siphluriscidae new fam- McCafferty, W. P. 1997. Ephemeroptera, pp. 89Ð117. In ily: Ephemeroptera). Fla. Entomol. 86: 345Ð352. R. W. Poole and P. Gentili [eds.], Nomina Insecta Nearc- tica, a check list of the Insects of North America. vol. 4: Received 24 March 2005; accepted 6 December 2005.