PHYLOGENY, CLASSIFICATION, MESOZOIC FOSSILS, AND BIOGEOGRAPHY OF THE LEIINAE (DIPTERA: MYCETOPHILIDAE)

SARAH SIQUEIRA OLIVEIRA Departamento de Ecologia

Instituto de Ciéncias Biologicas

Universidade Federal de Goids

DALTON DE SOUZA AMORIM Departamento de Biologia Faculdade de Filosofia, Ciéncias e Letras de Ribeirdo Preto Universidade de Sao Paulo

BULLETIN OF THE AMERICAN MUSEUM OF NATURAL HISTORY Number 446, 108 pp., 107 figures Issued March 16, 2021

Copyright © American Museum of Natural History 2021 ISSN 0003-0090

CONTENTS

PSDSETACE ess po Rae anat gst ah hea a setae al acouy BE tare ated anal gaee Bip ea Weta auedy a hate g ch anal gags 3 LiNtrOUUCH ORS as. ces coe SR Nah eae Sie ENDS. yiiat eh eet ER yt ely bine ER Br. oydalnt 3 Material eared VETO S: Ay SF suche Ss, Gea My MER ae essa Gate seek a each ac ccs eal, ae VR cease Sit oe 4 Material test agg: = eee te nih eye le teats leleane deta gaits Sita Meaialls crate Bian eee els 4 Preparation of specimens and morphology documentation. .......... 00... cece eee eee eee 5 PV POPC Vere COMSECUCH OI ee Oi i une ATO Nee SUE chen reat Ree en NNO: eat a 5 Character sampling and morphological terminology .......... 0... cece cece eee eee ee 6 RESUltsaCHLISGUSSIOM a ersga wets Macnee etal pe MRC hath and 32a ay atlge AEE nad Pee Gace te tla MEE sued 6 Relationships among subfamilies of Mycetophilidae.......... 0.0... eee eee eee eee eee 13 Monophy ser the Ueeiiniae shi ciea. ux sgn Beene megtoy ds het, x pbatscsumrrretedacdle Aaa esa raise Hee gS 18 ROGITCAOCT OTA sags fe sats: tan as tea Hit ps aah acane ema “thangs ah scar heann pee sce termsice tocueh adh anech pea thane oof 19 The backbone of thednennae phylogeny. nA. ee 08. wale aeds RECT og oe ORS Py bo a8 19

PeeP LD AACR aT Ve SV StETIILOT Nee LEM AS rata nth Sate Att a tonthls a2 SPARE lly RISO Anda hs oat Ta ea 28a 20 Selkirlsini EMG erlein We. secu osg aur, eines Sal ee Pig eta a, wore eat 20 WiesOphth ag lriniG inl eC). TOMS nr ated eco Rone, oe Pes esti Se Se eo ON ns Ae, oh id 2 21

Rom atitellimas tril: WON at. Aes An 2 hve a ee ay ee ochre ease ES An iy ee etka eee 23 CyclonewniniShaw. and Shaw a. Mats tn tle abe aaguiny Apt etyatnus Hem wie Stabe fut Mitabae eyo 4 24 PEATIOULTN Es Cl WATS cess te wotetiters eraevsiche abrasions ascend Maen aah tease ie eases ers rans besa rere WB osetia 24 Anomalomvainly AP WRMO V2. i wn. set Atle aN A in, ere anes AE. ah Ae Ale ARNE NE hn 27 Deiitin BO Wards siete. med ane nate «opts 0s agente Se = = AIR aan pao we ote a aes Sen aed eee =o 28 Male ternvinaliapatternsrirrthe Leiinae ny, fe ctte ts ac ea ye ues ey ance ae PN ees 33 Key lob the: Genet aro Leuiyaes i arenWeae AL ala ee nee ne. ne Enea Rea B Ay | Aiea) SBS Movi 54 mesazoic-tossil fecord_of the: Gelinas, fe ek. eacncte ee Bene eee ey eee Te oti epee a, oS 47 Alavamanota Blasodero. aide Avil lee We ain, re eee hi segue Remain Oe Foal oe Rane i ee oh Ps a es 58 BaisepestnOneisTia.D AC OUST ONAL Raper Ate k etes 2 alee SoA Mra gen ke RT Nes Bo a 59 DisparoleiaBlagoderor-and-Grinral dies. prea ns eter Ae Lee eI RL SR ee LL ee 60 Ecivepesthoneurasenderlein Stes claeis Be tera 2g eee eee eg as oa ily RP ey aaa 60 lie mola blasoderov. and Grim alt «fhe nok, t 8 weer cena ne eke Ae, caddy wn tk Recah aha, 60 UZietinasBlapaderoyZand: Ciritiial din, soa. Meets heey, eed te ce ge ocean Oy aC on hI 2 60 LecadeniiciaDlasodéeroy atid Grinaldis ie ioe matte et tyes ao aca tara ee arla soaps tema tara ot 61 Nedpeosia- Blavederon avd Grin alist. Stok eat ters ene. et BREE neh stra eee hu ate peed Sr 61 PULAP OA OCO SAT IVLGUIN Gm. eh et aR, AeA LM ek We STS Pike as os SRR NSCs sn la Skat ASS: 61 Protragoneura Blagoderov and. Grimaldiys a « 228< arma eas op48-4o 2 Mes i Pane a daw oe eee 62 lemaleia Blasoderoyeand. Grital dine: cay. pe puch ne eh ake METIS aca aoe ert ETN a 62 Zelinig Dias oderoveand Marital teeter ca h noe at reek WR fener OR RAEN Peis nee oer Le beews 2 62 bidgedsrap hic evolution-omthe: Lielitaes yw ect cree pA cts cw cetera le beers sregisn Ht tage 63 INCK MO WICGOMENIEG Mig ae ath sean et caten cues a's its at A amee e a re Peay oer el Ae ruta at ovis ca 66 RBFereICe sss 491542 bine saa pened d tad) nee AR AeA! he SPAN SAN PRR CA neh AA eae RL PE 67 Appendix A SLASE OPCharacterss jc. necnden hee eet ee Le E PRs ae ORME eh PEG je oh NE 73 Appendixes? Datartmattix, 2 sto nks eg, Mp ete ee ee a nN. hee ial Rh epee ahi eg 77 APpendixss cistoh Matehalexanniticds, oe Monn eer cn eae unk ee, Seti tA 84 Appendix 4. Mycetophilid fossils (modified from Evenhuis, 2014) assignable to subfamilies. 100 Appendix 5. Age and location of mycetophilid amber and compression fossil sites......... 108

ABSTRACT

The relationships among the genera of fungus gnats in the mycetophilid subfamily Leiinae are unclear, and the monophyly of this group is questionable. This monograph provides an extensive phylogenetic study of the Leiinae based on morphological data from a large taxonomic sample, including all genera that have ever been assigned to the subfamily and a wide outgroup sampling to properly test subfamilial monophyly. A data matrix with 128 morphological features of 117 terminal taxa was carried out under parsimony using different implicit weight schemes. All recov- ered topologies support a monophyletic Leiinae that is more restricted than the usual delimitation of the subfamily. We found no consistent evidence that a clade with Docosia Winnertz, Novakia Strobl, Ectrepesthoneura Enderlein, and Tetragoneura Winnertz group together with the remaining genera of Leiinae. A name with subfamily rank—Tetragoneurinae, already present in the litera- ture—is used here to refer to this group. The allactoneurine genera Sticholeia Soli and Allactoneura de Meijere form a clade with the leiine genus Leiella and the genera of Manotinae, which is deeply nested within the Leiinae. The male terminalia patterns found within the subfamily are analyzed and illustrated. A classification for the Leiinae is proposed grouping 33 genera in seven clades ranked as tribes: Selkirkiini Enderlein, Megophthalmidiini, trib. nov., Rondaniellini, trib. nov., Cycloneurini Shaw and Shaw, Manotini Edwards, Anomalomyiini, trib. nov., and Leiini Edwards. A key for the world genera of Leiinae is also provided. The Cretaceous mycetophilid fossil record

is revisited and the biogeographic evolution of the Leiinae is discussed.

INTRODUCTION

Fungus gnats of the family Mycetophilidae have immature stages mostly associated with the fruiting bodies, hyphae or spores of fungi. The Mycetophilidae are the second most species-rich family of the suborder Bibionomorpha (see Amorim and Yeates, 2006), with 233 genera and about 4500 species, described from all biogeo- graphic regions (Pape et al., 2011), second in number of species only to the Cecidomyiidae. They are known in the fossil record from the Cretaceous through the Cenozoic, where they are diverse and sometimes abundant (Amorim and Silva, 2002; Blagoderov and Grimaldi, 2004; Evenhuis, 2014). The family is clearly monophy- letic (e.g., Soli, 1997; Rindal et al., 2009a) and often divided in the subfamilies Sciophilinae, Gnoristinae, Mycomyiinae, Leiinae, Manotinae, Allactoneurinae, and Mycetophilinae (Tuomikoski, 1966; Hennig, 1973; Vaisanen, 1984; Matile, 1989; Rindal et al., 2009a).

Phylogenies have been published for the mycetophilid subfamilies Manotinae (Hippa et al., 2005), Mycetophilinae (Rindal and Soli, 2006; Rindal et al., 2007, 2009b), Sciophilinae (Borkent

and Wheeler, 2013), Gnoristinae, and Myco- myinae (Kasprak et al., 2019), based on morpho- logical and/or molecular information. The Leiinae have so far not shown up on the phylo- genetic radar.

The composition of the Leiinae accepted by most authors comprises 37 genera and about 550 species worldwide (Oliveira and Amorim, 2012). There are 54 species of the subfamily known from fossils, which include 12 additional extinct genera, eight of which are in Cretaceous amber (Blagoderov, 1998a, 1998b, 2000; Blagoderov and Grimaldi, 2004; Evenhuis, 2014).

A tribal rank for the Leiinae was originally proposed by Edwards (1925), who established that a short R, (usually shorter than r-m) and a longitudinal r-m aligned with the second section of Rs would be diagnostic for the group. Edwards (1925) himself, however, pointed out that there are some exceptions for these features, e.g., Ron- daniella Johannsen, Docosia, and Tetragoneura.

Hendel (1936) gave subfamily rank to the Leiini, but the generic composition and the diagnosis of the group have been repeatedly questioned (Tuomikoski, 1966; Hennig, 1973; Sali, 1997; Soli et al., 2000; Hippa et al., 2005;

4 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY

Jaschhof and Kallweit, 2009). Tozoni (1998) recovered a monophyletic Leiinae, supported by the reduction of the length of R;, the first sec- tion of Rs nearly transverse, R, missing, and an incomplete mediopleural suture, which is not produced on its lower fourth. The taxon sam- pling of studies of the phylogenetic relation- ships among mycetophilids in general (e.g., Soli, 1997; Tozoni, 1998; Hippa et al., 2005; Rindal et al., 2009a; Sevéik et al., 2013), how- ever, has been considerably limited and none of these studies had a wide sampling of leiine genera.

A proper test for the monophyly of the Leiinae and establishing the relationships among its gen- era to provide a robust classification for the sub- family is entirely dependent on: (1) a wide sampling of the genera of the subfamily; and (2) a proper choice of outgroups to have a reliable test of its monophyly. This paper conducts a for- mal phylogenetic analysis of the Leiinae based on morphological information of 117 terminal taxa—all genera currently in the subfamily, all extant genera that may have been referred to as possibly connected to the leiines and a large number of outgroups, including allactoneurines and manotines.

MATERIAL AND METHODS MATERIAL

Specimens used in our study were obtained from the following collections (including acro- nyms used in the text):

AMSA Australian Museum, Sydney, Australia

ANIC Australian National Insect Collection, Canberra, Australia

CEUA Coleccién de Entomologia of the Univer- sity of Antioquia

CNC Canadian National Collection of Arach- nids, Nematodes and Insects, Ottawa, Canada

DZUP Colecao de Entomologia Padre Jesus Santiago Moure da Universidade Federal do Parana, Curitiba, Brazil

NO. 446

FMNH Finnish Museum of Natural History, Zoological Museum, University of Hel- sinki, Helsink, Finland

IAvH Instituto de Investigacidén de Recursos Biolégicos Alexander von Humboldt, Bogota, Colombia

INPA Instituto Nacional de Pesquisas Amaz6ni-

cas, Manaus, Brazil

LMED Laboratorio de Morfologia e Evolugao de Diptera, Faculdade de Filosofia, Ciéncias e Letras de Ribeirao Preto da Universi- dade de Sao Paulo, Ribeirao Preto, Brazil

MNHN Muséum National d'Histoire Naturelle, Paris, France

MZUSP Museu de Zoologia da Universidade de Sao Paulo, Sao Paulo, Brazil

NHM Natural History Museum, London, United Kingdom

NMSA Kwa-Zulu-Natal Museum South Africa, Pietermaritzburg, South Africa

NZAC New Zealand Arthropod Collection, Auckland, New Zealand

SAMC Iziko South Africa Musuem, Cape Town, South Africa

SMOC Silesian Museum, Opava, Czech Repub- lic

Properly verifying the monophyly of the Leiinae requires a wide sampling of genera that at any time have been assumed to be connected with the subfamily. Particularly, Tetragoneura and allied genera (as Novakia, Ectrepesthoneura, and Docosia) have been accepted either as lei- ines, gnoristines or as an independent group. The initial delimitation of an ingroup for the analysis here included 95 species of all 37 “Leiinae s.l” genera (including Tetragoneura, Ectrepesthoneura, Novakia, and Docosia). Whenever possible, we tried to use the type species of each leiine genus in the analysis. The genera Allactoneura and Sticholeia have often been placed in a subfamily of their own, but their relationship to the leiines (see discussion below) has been stressed by different authors. The fact that the manotines have often been associated with the Allactoneurinae also makes

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE) 5

it indispensable that all of its genera should be integrated into the analysis.

Outgroup sampling is a key issue, since there is no consensus in the literature about the posi- tion of the Leiinae in the phylogeny of the Myce- tophilidae. Our outgroup list includes nine species of five genera of Sciophilinae, six species of six genera of Gnoristinae, two species of two genera of Mycomyiinae, and four species of four genera of Mycetophilinae (two Exechiini and two Mycetophilini). One species of Keroplatidae was used to root the entire tree. The full matrix includes 117 terminal taxa. Complete informa- tion of the specimens used in this study is included in the appendix 3. ‘The list of characters is in appendix 1 and the data matrix is in the appendix 2. A complete list of known Myceto- philidae fossils (appendix 4) and their fossil deposits (appendix 5) were used to infer the age of the main nodes of the backbone of the Leiinae phylogeny. Over a hundred additional species of mycetophilids were slide-mounted and studied, although not formally included in the matrix.

PREPARATION OF SPECIMENS, MORPHOLOGY DOCUMENTATION, AND ABBREVIATIONS

When available, both males and females of each species were studied. Most specimens were dissected and mounted on permanent slides. Specimens were cleared with KOH, dehydrated in ethanol, and mounted in Canada balsam (modified from Walker and Crosby, 1988; Huber and Reis, 2011). In some cases, after clearing, the terminalia were studied in temporary slide mounting with glycerine or gelatin with phenol (modified from Zandler, 2003).

The habitus of the specimens and morpho- logical details of the structures were studied using light microscopy and were photographed with a Leica DC500 camera attached to a Leica stereomicroscope model MZ-16 or a com- pound microscope model Leica DM2500. Pho- tos were stacked with Helicon Focus 6. The morphological structures were drawn using a camera lucida attached to the compound

microscope. Images were edited with Adobe Photoshop CC. All terminal taxa had speci- mens studied except of the fossils species and the genus Paramanota. Data for Paramanota in the matrix were taken from the literature except for the wing, obtained from a photo- graph kindly made available by Jan Sevéik.

Along the discussion of male terminalia pat- terns in the Leiinae, we refer to published illus- trations for most genera. Some leiine genera do not have any published illustrations of male terminalia. We include here stacking photo- graphs of 27 species of 20 genera in the subfam- ily. Slide mounts show relatively transparent structures at different focus levels and stacking does not work as with pinned specimens: struc- tures at different levels often blur together. Our photographs provide illustrations of the general pattern of the male terminalia of part of the lei- ine genera and we refer to illustrations as they appear published on paper. A full study of the details of the male terminalia morphology in each genus or species, however, is beyond the scope of this paper. Abbreviations for male ter- minalia plates as follows: adlgc, apico-dorsal lobe of gonocoxite; allgc, apico-lateral lobe of gonocoxite; avlgc, apico-ventral lobe of gono- coxite; aed, aedeagus; allgc, apico-lateral lobe of gonocoxite; avlgc, apico-ventral lobe of gonocoxite; cerc, cercus; ej ap, ejaculatory apodeme; epand, epandrium; gonocx, gono- coxite; gonocx apod, gonocoxite apodeme; gonst, gonostylus; gsdl, gonostylus dorsal lobe; gsl, gonostylus main lobe; gsml, gonostylus medial lobe; gsvl, gonostylus ventral lobe; hypd, hypandrium; Idlep, laterodistal lobe of epandrium; pm, paramare; pm apod, parameral apodeme; st9, sternite 9; syngcxm, syngono- coxite medial sclerite; teg, tegmen.

PHYLOGENY RECONSTRUCTION The character matrix was constructed using

WinClada (version 1.89). Characters were treated as unordered; unobserved states and inapplicable

6 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY

(<9 >

data were coded respectively as “?” and “-- Some characters were coded as absent or present, in some cases causing interdependence. We retain these characters as separate in order to extract pertinent phylogenetic data from the morphological differences we observed (Lee and Bryant, 1999; Strong and Lipscomb, 1999).

The phylogenetic analyses of the matrix were made under Fitch parsimony (1971), imple- mented using TNT (Tree Analysis Using New Technologies—Willi Hennig Society Edition; Goloboff et al., 2008). Topologies in TNT were obtained using New Search Technology (Golo- boff, 1999; Nixon, 1999; Goloboff et al., 2008), recommended for matrices with more than 100 terminals. According to Goloboff (1999) and Nixon (personal commun.), the new technolo- gies should be used together; Drifting and Ratchet are very similar and the best method for complex data sets is Ratchet (Nixon, 1999). The following parameters were used for the analy- ses: Max. trees 10,000; Random seed 0; Random addition sequences 200, Sectorial search (sect: slack7); Ratchet 200 interactions; Tree fusing 5 cycles.

The rooting procedure followed Nixon and Carpenter (1993) using an unequivocal out- group, in this case a species of Keroplatidae. Final trees files were obtained using WinClada software, edited in Adobe Illustrator CC. Bremer support (Bremer, 1994) was calculated for the strict consensus tree using TNT to indicate the extra steps required to collapse a branch. Subop- timal trees with 1-20 extra steps with TBR (Tree Bissection Reconnection) were used to calculate Bremer support values.

We used implied weighting schemes to reduce the potential influence of incongruent characters over nested characters (Goloboff, 1993). In other words, properties of the data were used to reduce the chances that random association between incongruent characters outperform nested char- acters under equal weight. Initial analyses of the data matrix were made in TNT under different k values—between 1 and 10, 15, 20, and 25—as well as an analysis with equal weight to assess its

NO. 446

effect on the final topology. A tree was also obtained using the script “setk.run” (available from Salvador Arias, unpublished data, to choose the best k value) with TNT based on our dataset, which resulted in k = 24.22175. The tree used to discuss character evolution was the majority consensus of the equal weight analysis.

CHARACTER SAMPLING AND MORPHOLOGICAL TERMINOLOGY

The matrix (appendix 2) has morphological characters of male and female adults. Some of the characters used here were proposed in the phylogenetic analyses of the Mycetophilidae by Sali (1997), Tozoni (1998), Rindal and Soli (2006), Amorim and Rindal (2007), and Borkent and Wheeler (2013). Several characters are proposed here for the first time. The mor- phological terminology follows Cumming and Wood (2017), while structures particularly of the thorax and male terminalia features follow Sali (1997), Amorim and Rindal (2007), and Matile (1990). We use here the term “spines,” in accordance with Cumming and Wood (2017), for hardly sclerotized bristles. Unnamed clades on the phylogeny are referred to using the group+ artifact (Amorim, 1982), in which, e.g., the group (A + (B + (C + (D +E)))) is shortened to “group-A*,’ i.e., the clade including A plus its sister group.

RESULTS AND DISCUSSION

This is the first cladistic study of the Leiinae with a complete generic sampling and a sub- stantial number of characters. The need for a study of the Leiinae with a comprehensive sam- pling was made clear in the literature (e.g., Jas- chhof and Kallweit, 2009). Our study includes a wider sampling within some of the more speci- ose genera to address the question of their monophyly. The generic sampling outside the Leiinae was particularly designed to test of the monophyly and, hence, the generic composition of the subfamily.

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE)

Neoallocotocera fusc

Tasmanina gracilis

5

=—

—=

<i

Aneura sp.

Parvicellula sp.

FIGS. 1-5. Wings of Sciophilinae species. 1. Neoallocotocera fusca Tonnoir. 2. Aneura sp. 3. Tasmanina gracilis

Tonnoir. 4. Eudicrana splendens Lane. 5. Parvicellula sp.

A series of plates with the wings of all Leiinae genera were included here (figs. 1-63), for ease in following the wing characters in the list of characters and for using the key to the genera of the subfamily.

The data matrix (appendix 2) has a total of 128 characters (73 binary and 55 multistate), of which 34 are from head structures, 43 from tho- rax (including legs), 43 from wing, and 8 from male terminalia (appendix 1). In most cases, the state “O” already corresponds to the most plesio- morphic condition found within the Mycetophi- lidae. The analysis under equal weights resulted in 119 most-parsimonious trees, which majority consensus is in figure 96 and the strict consensus is in figure 97. The consistency index for the

majority consensus tree is 0.19, while the reten- tion index is 0.76, with 1,132 steps. Figure 97 shows the Bremer support for all nodes. The implied weight analysis under k=3 resulted in one most parsimonious tree (fig. 98), while the script “setk.run” k = 24.22175 results in a single most parsimonious tree (fig. 99). Both these trees are very similar in topology to the majority con- sensus tree of the equal weight analysis.

We used the majority consensus for the dis- cussion, since it provides slightly more informa- tion in a phylogenetic framework. The majority consensus keeps some of the clades not present in the strict consensus, which have been justified in the literature. We also carefully considered the differences between the tree topology of the tree

BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 446

Palaeodocosia vittata Boletina aS

es ec ee

es ee eria sp.

14 Mycomya sp.

Rymosia sp. Exechiopsis sp.

Epicypta sp.

Mycetophila sp.

FIGS. 6-17. Wings of Gnoristinae, Mycomyinae, and Mycetophilinae species. 6. Palaeodocosia vittata (Coquil- lett). 7. Boletina obscura Johannsen. 8. Schnusea caiabii Lane. 9. Dziedzickia metallica Lane. 10. Austrosynapha hirta Tonnoir. 11. Synapha sp. 12. Mycomya sp. 13. Neoempheria sp. 14. Rymosia sp. 15. Exechiopsis sp. 16.

Mycetophila sp. 17. Epicypta sp.

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE)

Ectrepesthoneura colyeri

Tetragoneura borgmeieri

FIGS. 18-21. Wings of tetragoneurine species. 18. Docosia sciarina (Meigen). 19. Novakia miloi Kerr. 20. Ectrepesthoneura colyeri Chandler. 21. Tetragoneura borgmeieri Edwards.

in figure 96 and the trees obtained with k = 3 (fig. 98) and with k = 24.22175 (fig. 99).

Phylogenies are complex reconstructions that integrate into a single tree a set of individual hypotheses about relationships between the termi- nals (i.e., hypotheses on smaller clades). Nested subgroups of hypotheses and mutually indepen- dent hypotheses are present in any cladogram—e.g., the potential paraphyly of Mycetophila Meigen does not contradict a hypothesis of monophyly of the Mycetophilinae. The robustness of each node, hence, is often not affected by the weakness or robustness of clades in other parts of the tree. The assessment of the reliability of different clades in a phylogeny should be made case by case while considering their respective hypotheses.

The consistency index in the tree is relatively low (0.19), expressing the relatively high charac- ter plasticity. The retention index, however, is relatively high (0.75), indicating that incongru- ent characters are not significantly affecting the backbone of the tree. That explains the consider- ably good values for the Bremer support for most larger clades within the Leiinae tree (fig. 97).

Four names of the seven taxa with tribal rank in our classification (fig. 100) were already pro- posed in the literature (Manotini Edwards, 1925; Leiini Edwards, 1925; Selkirkiini Enderlein, 1940; Cycloneurini Shaw and Shaw, 1951). Each of the tribes is considered in detail in the discus- sion below, and we provide a formal diagnosis for each tribe. The analytical procedures used here to deal with the data matrix, with different k values for weighting schemes, allows spotting the genera that change their position in topolo- gies with different parameters (i.e., different k values). Instead of considering as correct the position of these rogue genera in any particular tree (and to reflect it in the classification), we preferred to keep them unplaced in our tribal classification of the Leiinae.

Very few male terminalia characters were included in the analysis. The reason is that gathering male terminalia information at this stage for all terminals would result in a matrix with a high proportion of missing data—due to noncomparable features, to access to infor- mation and to unsolved homology issues. Our

10 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 446

wi nbcatranatel ot

Garretella shermani Paraleia nubilipennis

pag nn

Thoracothropis cypriformis Gracilileia redunda

Mohelia matilei

30

Aphrastomyia shannoni

Paracycloneura apicalis

FIGS. 22-30. Wings of Leiinae species of Selkirkiini, Megophthalmiini and rogue genera. 22. Garretella sher- mani (Garrett). 23. Paraleia nubilipennis (Walker). 24. Thoracothropis cypriformis Freeman. 25. Gracilileia redunda Matile. 26. Trichoterga monticola Tonnoir and Edwards. 27. Megophthalmidia nigra Freeman. 28. Mohelia matilei Oliveira. 29. Aphrastomyia shannoni Lane. 30. Paracycloneura apicalis Tonnoir and Edwards.

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE) 11

Indoleia bisetosa i ae Rondaniella dimidiata

Procycloneura sp.

Tonnwardsia aberrans

FIGS. 31-40. Wings of Leiinae species of Rondaniellini and Cycloneurini. 31. Indoleia bisetosa (Edwards). 32. Rondaniella dimidiata (Meigen). 33. Waipapamyia elongata Jaschhof and Kallweit. 34. Cawthronia nigra 'Ton- noir. 35. Sigmoleia melanoxantha Tonnoir and Edwards. 36. Paradoxa paradoxa Jaschhof. 37. Paradoxa fusca Marshall. 38. Cycloneura flava Marshall. 39. Tonnwardsia aberrans (Tonnoir). 40. Procycloneura sp.

12 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY

NO. 446

Leiella ochreocalcar

Manota sp.

Promanota malaisei

Eumanotasp.

Sele i Si Lalla = bean, CaS = a i ay *

Paramanota furcillata

FIGS. 41-48. Wings of Leiinae species of Manotini. 41. Leiella zonalis Edwards. 42. Leiella ochreocalcar Ender- lein. 43. Sticholeia cheesmanae Soli. 44. Allactoneura papuensis Bechev. 45. Manota sp. 46. Promanota malaisei Tuomikoski. 47. Eumanota sp. 48. Paramanota furcillata Hippa.

efforts during the early stages of this analysis with a larger number of male terminalia char- acters were not encouraging. Missing data have a damaging effect on phylogenetic analy- ses, with loss of information. We preferred instead to have a section in the paper to address specifically male terminalia patterns in the Leiinae. Since most of the characters cor- respond to features that define male terminalia

patterns at the generic level, sometimes below the level of genus, the decision does not affect much the backbone of the phylogeny.

The same approach applies to the presence of fossils as terminals. There is no chance to avoid large amounts of missing data in the matrix while including fossils in the data matrix. Again, our attempts at earlier stages of this study to include fossils in the matrix

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE) 13

Anomalomyia guttata

51

Ateleia spadicithorax

Acrodicrania angustifurca

FIGS. 49-51. Wings of Leiinae species of Anomalomyiini. 49. Anomalomyia guttata (Hutton). 50. Ateleia spadicithorax Skuse. 51. Acrodicrania angustifurca Skuse.

resulted in polytomies, losing information even at the level of clades with subfamily rank. The fossil genera, hence, are discussed one by one ahead in a separate section, comparing the features available in the descriptions to the characters in the analysis.

RELATIONSHIPS AMONG SUBFAMILIES OF MYCETOPHILIDAE

In the majority consensus (but not in the strict consensus) tree (fig. 96), the group of scio- philine genera sampled here forms a single clade. A monophyletic sciophiline was obtained by Borkent and Wheeler (2013), but Sali (1997: 49) found that Paratinia Mik and Drepanocercus Vockeroth do not comprise a monophyletic group with the remaining sciophilines. We do not have Paratinia and Drepanocercus in our taxon sampling and, hence, our analysis does not conflict with or confirm Soli’s (1997) or Borkent and Wheeler’s (2013) conclusions about the monophyly of the Sciophilinae. There is, how- ever, a large core group of sciophiline genera that comprise a well-defined clade, as stated by Soli (1997) and Borkent and Wheeler (2013).

Plesiomorphies have been often used as diag- nostic features of some of the mycetophilid sub- families, resulting in confusion over the position of some genera in the system. Doubts have been repeatedly raised particularly about the mono- phyly of the Gnoristinae and about its position in the phylogeny of the mycetophilids (e.g., Vaisanen, 1986; Soli, 1997; Soli et al., 2000; Rindal and Soli, 2006; Jaschhof and Kallweit, 2009). It should be no surprise, then, that, even with our limited sampling of gnoristines, the genera of the subfamily fit into two separate clades in our tree, one of them closer to the clade (Mycomyinae + Mycetophilinae) than the other. Borkent and Wheeler’s (2013) phylogeny of the Sciophilinae is rooted in Mycomya Rondani, so their result cannot be used for the relationships among mycetophilid subfamilies. All trees in Seli’s (1997) study also show the gnoristines as paraphyletic. Kasprak et al. (2019) have the gnoristines as a grade (i.e., a paraphyletic group) within which the mycetophilines are nested.

The position of the Mycomyinae as sister to the Mycetophilinae (fig. 101) was also recovered by Rindal and Soli (2006) based both on morpho- logical and molecular data. In their study, Manota

14 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY NO. 446

Leia ventralis

Neoclastobasis kamijoi Greenomyia stackelbergi

Leia arsona Leia winthemi

60

Clastobasis vicina 63 Leia spinifera

Clastobasis sp. Leia amapaensis

FIGS. 52-63. Wings of Leiinae species of Leiini s.s. 52. Caledonileia pusilla Matile. 53. Leia ventralis Say, with a teratology, M4 missing. 54. Neoclastobasis kamijoi (Sasakawa). 55. Greenomyia stackelbergi Zaitzev. 56. Leia fascipennis Meigen. 57. Clastobasis alternans (Winnertz). 58. Leia arsona Hutson. 59. Leia winthemi Lehmann. 60. Clastobasis vicina Matile. 61. Leia spinifera Edwards. 62. Clastobasis sp. 63. Leia amapaensis Lane.

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE) 15

Williston and Tetragoneura behave as rogue taxa, either close to the base of the family or nested within the sciophilines. The possible paraphyly of the Gnoristinae and the monophyly of a clade (Mycomyinae + Mycetophilinae) are beyond the scope of this study, but it is interesting that the results here are consistent with Rindal and Solis (2006) study based on very different matrices. The monophyly of a clade including Mycomyinae, Mycetophilinae, and a paraphyletic Gnoristinae was also found by Kaspyak et al. (2019) using molecular data, although their tree has the Myco- myinae as sister of the “Gnoristinae” plus Myceto- philinae. In their study, they sampled two species of manotines that compose a grade at the base of the Mycetophilidae phylogeny.

Kaspfrak et al. (2019) obtained Allactoneura as the sister of Leia Meigen, both comprising together the sister clade of Garretella—these are the only leiines sampled in their study. The posi- tion of both manotine species in their tree greatly differs from what was found here with a wider sampling of manotine and nonmanotine leiine genera and of mycetophilids of other subfamilies. Their results also disagree with the phylogeny of the Exechiini (Burdikova et al., 2019) obtained with molecular data, in which, among the out- groups, the sciophilines compose a grade at the base of the mycetophilids, with one species of Manota coming out as sister of a clade including (Rondaniella + Leia) and (Mycomyiinae + Gnoristinae + Mycetophilinae). Finally, the results from Kasprak et al. (2019) also disagrees from the reconstruction from Sevéik et al. (2013), in which all four manotine genera group in a clade with the remaining sampled leiine genera (1.00 posterior probability). The node with the sciophilines joining gnoristines +mycomyiines in Sevéik et al’s (2013) paper has low support (0.68 posterior probability). In that study, as was found here, Ectrepesthoneura, Docosia, and Novakia do not group with the leiines, but with the gnoris- tine and mycomyiines (no mycetophilines were included in their analysis).

The question of the monophyly of the Leiinae and its position in the system is the core of this

paper. Edwards (1925) commented on the similar- ities between Tetragoneura and Ectrepesthoneura, keeping both genera in the Leiinae. This position was later followed by Hackman et al. (1988), Soli (1997), and Kurina (2004). Tuomikoski (1966) mentioned that both these genera should be excluded from the Leiinae, placing them with Synapha Meigen in the Gnoristinae. Vaisanen (1986) placed Tetragoneura and Ectrepesthoneura in the Gnoristinae, but retained Docosia within the Leiinae, a position also held by Bechev (2000). Chandler (1994), Chandler and Blasco-Zumeta (2001), Chandler (2004), and Chandler et al. (2006) kept Novakia and Docosia in the Leiinae, while Chandler (2004) and Chandler et al. (2006) have Tetragoneura and Ectrepesthoneura in the Gnoristinae. In Tozoni’s (1998) phylogenetic study of the family, Ectrepesthoneura is the sister group of Novakia, inside a clade also including Tetrago- neura, Trichoterga Tonnoir and Edwards, Aphras- tomyia Coher and Lane, Thoracotropis Freeman, Impleta Plasmann, and Docosia. Jaschhof and Kallweit (2009) also proposed that Tetragoneura and Novakia (with some other genera) would have gnoristine affinities. The sampling of Gnoristinae genera in this study is relatively small (6 of 29 genera) and the question of the monophyly of the Gnoristinae still needs proper scrutiny.

The position of the Tetragoneura group of genera in the phylogeny of the Mycetophilidae is pending, but not the monophyly of this group nor its position outside the Leiinae. There are several apomorphic features—characters 29:1, 35:1, 63:2, 74:1, 87:1, 93:1—supporting the clade (Docosia + (Novakia + (Ectrepesthoneura + Tetragoneura))), with a Bremer support of 2. The position of Novakia nested within the group has a Bremer support of 4 and corroborates many views in the literature about its relationships with other genera—in fact, the wing venation of these genera is considerably similar (figs. 18-21). Moreover, several features support that Ectrep- esthoneura and Tetragoneura are sister genera.

Regarding the position of this clade in the evolution of mycetophilids, we could not find definite evidence that (Docosia + (Novakia +

16 BULLETIN AMERICAN MUSEUM OF NATURAL HISTORY

(Ectrepesthoneura + Tetragoneura))) would be sister to the core leiines—although this position cannot be entirely excluded. In none of our trees, however, does this clade nest within the Leiinae. Both the majority consensus (fig. 96) and the strict consensus trees (fig. 97) show these four genera composing a clade in a polytomy with the Leiinae and the clade (Gnoristinae + Myco- myinae + Mycetophilidae). Indeed, one of the possible solutions for this trichotomy is with the clade with Tetragoneura as sister of the remain- ing leiines, but this is not our conclusion with the currently available data.

This position of the Tetragoneura group raises the problem of its status within the Mycetophilidae. Meunier (1900) proposed a taxon of subfamily rank—Tetragoneurinae—that applies to this clade (see Sabrosky, 1999). On the one hand, if Tetrago- neura and related genera collectively correspond to a leiine subclade, we would have to follow Vock- eroth (1981), who showed that the name Tetrago- neurinae has priority over Leiinae—proposed by Edwards (1925). On the other hand, if this small clade is sister to the Leiinae or sister to a larger clade that includes two or more subfamilies, as appears in our results, it can have subfamily status, separate from the remaining mycetophilid subfami- lies, which is the position taken here.

In dealing with the Cycloneura group, Jas- chhof and Kallweit (2009) advocated that the problem of the Leiinae is broader and that a proper analysis should encompass additional genera. They stated that the two characters described by Edwards (1925) to delimit the lei- ines—short Rj, usually shorter than the length of r-m and a longitudinal r-m, aligned with the sec- ond sector of Rs—were solid enough to delimit the group for the genera known at Edwards's time, but we now know genera that do not prop- erly fit into this definition. However, Sigmoleia Tonnoir and Edwards (fig. 35), in one hand, has R, longer than r-m and r-m is not aligned to R. On the other hand, an elongate r-m aligned with the second sector of Rs is present in tetragoneu- rine genera (and in some degree also seen in the Exechiini mycetophilines).

NO. 446

In Jaschhof and Kallweit’s (2009) opinion, genera such as Aphrastomyia (fig. 29), Gracilileia Matile (fig. 25), Mohelia Matile (fig. 28), Nova- kia (fig. 19), and Tetragoneura (fig. 21) should be excluded from the “Leiini’” In most of these genera, Sc generally “ends in R” (not in C)— actually, the tip of Sc beyond sc-r is lost, so Sc continues through sc-r to reach bR. This feature is typically seen in Gnoristinae (although present elsewhere). Our analysis supports their view on Novakia and Tetragoneura.

The position of Allactoneura (fig. 44) and Sticholeia (fig. 43) deeply nested within the Leiinae (together with the Manotinae) should not at all be a surprise. Edwards's (1925) original placement for Allactoneura was actually as a manotine. Shaw and Shaw (1951) understood that Allactoneura shares similarities with Procy- cloneura Edwards, especially in the thoracic pleura, assuming leiine affinities for the genus. This position was clearly defended later by Tuomikoski (1966), who considered the genus a member of the Leiinae. Zaitzev’s (1982a: 912) revision of Allactoneura indicated that the genus is “sufficiently isolated from representatives of the tribe Leiini both by a whole complex of char- acters of the imago and of the larva,’ but con- cludes that “judging by the figure of the wing venation (Johannsen, 1909) and the structure of the thoracic sclerites (Shaw and Shaw, 1951), the genus Allactoneura is apparently close to the New Zealand genus Cycloneura Marshall” This shows that Zaitzev (1982a) probably had a slightly more restrictive concept of the Leiinae (possibly with Leia and more close allies), but he understood that Allactoneura belongs to a wider leiine arrangement. Matile (1993) accepted Allac- toneura as part of the “Leiini s.1.”

In Soli’s (1997: fig. 45) phylogeny of the Myce- tophilidae, obtained with majority consensus, Leia and Rondaniella come out together sister to Allactoneura, the clade with these three genera sister to Eumanota Edwards. This leeine clade is sister to the genera of Mycetophilinae. This led Soli (1997) to reject the “Allactoneurini” as pro- posed by Vaisanen (1986). When Sali (1996: 4)

2021 OLIVEIRA AND AMORIM: THE LEIINAE (DIPTERA: MYCETOPHILIDAE) i

described Sticholeia, he specifically assigned the genus to the Leiinae, mentioning that “the com- bination of strong, recurved bristles behind the eyes and a regular arrangement of the tibial trichia makes Sticholeia key out as Eumanota (subfamily Manotinae) in most available keys.” He also stated that Sticholeia has “a combination of characters found in members of the subfamily Manotinae and in Allactoneura de Meijere, 1907, and some other genera in the tribe Leiini [s.1.] Soli (1996: 10) added further ahead, “like Allac- toneura and Leiella, Sticholeia has a very short stem of the median fork and a costa not pro- duced beyond the tip of R,,;. In Allactoneura and Leiella, the abdomen is densely clothed by scale- like setae, a character not present in other groups of mycetophilids. Available evidence thus sug- gests that Sticholeia is the sister group of Allacto- neura and Leiella Enderlein combined,’

Jaschhof and Kallweit (2009) extensively dis- cussed features shared by Allactoneura and Sticholeia with other leiines (especially Leiella and Procycloneura), and considered Allactoneura “properly placed” within the leiines. Finally, in Sevcik et al’s (2013) molecular phylogeny of mycetophilids, dealing with a limited taxon sam- pling, the Manotinae are sister to a clade with species of Leiinae (including only Leia and Clas- tobasis Skuse) mixed with Allactoneura and Sticholeia—mycomyines and mycetophilines were not part of the analysis.

We consistently found a clade in which Sticholeia is sister to (Allactoneura + (Mano- tinae)) deeply nested within the Leiinae. A sim- ilar conclusion emerges from Hippa et als (2005) study, as considered below. Apparently, the distinctiveness of Manota and Allactoneura is the main reason for these two genera to have been placed in a separate subfamily. The conse- quence, however, was that, accepting Manotinae and Allactoneurinae as separate subfamilies, plesiomorphies had inevitably to be used as diagnostic features for the Leiinae, correspond- ing to a paraphyletic leiine.

Manota is at the core of this discussion. The wing venation of the genus is rather highly mod-

ified compared to other mycetophilids (fig. 45), while the other three manotine genera— Eumanota, Promanota Tuomikoski, and Para- manota Tuomikoski (respectively figs. 47, 46, and 48)—are much less derived. Edwards (1933) and later Sevéik et al. (2013) clearly stressed that Eumanota forms “a transition between Mano- tinae and Leiinae” (Sevéik et al., 2013: 4). Hippa et al’s (2005) analysis established the relation- ships among the genera of Manotinae. Their sampling of nonleiine genera was intended to root their analysis of the phylogeny of the Mano- tinae, not to recover the position of the mano- tines within the mycetophilids. Our analysis recovers exactly the same results for the relation- ships among the manotine genera obtained by Hippa et al. (2005), but it is conceivable that Pro- manota could be sister to Manota. In their analy- sis, Procycloneura is sister to the “Manotinae,’ while the other sampled leiine genera fit in two other clades. One of these clades has Ectrepestho- neura, Aphrastomyia, and Mohelia in a clade sis- ter to (Procycloneura + Manotinae). The other clade has Mycetophila as sister to a clade with leiines including Leiella and Rondaniella together sister of (Leia + Greenomyia Brunetti) and (Allactoneura + Sticholeia).

In our results, the clade including allactoneu- rines and manotines, as mentioned above, is deeply nested within the Leiinae. This appears consistently in trees obtained with all weighting schemes. The lack in the literature of a formal phylogenetic analysis of the Leiinae with wide taxon sampling and a proper selection of out- groups is probably behind the decision of many authors to keep the Allactoneurinae and the Manotinae separate from the Leiinae, despite evidence of the leiine-manotine-allactoneurine connection. We here ranked the clade of allacto- neurines and manotines as a tribe within Leiinae.

Matile (1978) referred to groups of genera