Multi-Gene Phylogenies Indicate Ascomal Wall Morphology Is a Better

Molecular Phylogenetics and Evolution 35 (2005) 60–75 www.elsevier.com/locate/ympev

Multi-gene phylogenies indicate ascomal wall morphology is a better predictor of phylogenetic relationships than ascospore morphology in the Sordariales (Ascomycota, Fungi)

Andrew N. Miller a,¤, Sabine M. Huhndorf b

a Illinois Natural History Survey, Center for Biodiversity, 607 E. Peabody Dr., Champaign, IL 61820, USA b The Field Museum of Natural History, Botany Department, 1400 S. Lake Shore Dr., Chicago, IL 60605-2496, USA

Received 3 December 2003; revised 20 October 2004

Abstract

Ascospore characters have commonly been used for distinguishing ascomycete taxa, while ascomal wall characters have received little attention. Although taxa in the Sordariales possess a wide range of variation in their ascomal walls and ascospores, genera have traditionally been delimited based on diVerences in their ascospore morphology. Phylogenetic relationships of multiple representatives from each of several genera representing the range in ascomal wall and ascospore morphologies in the Sordariales were estimated using partial nuclear DNA sequences from the 28S ribosomal large subunit (LSU), -tubulin, and ribosomal polymerase II subunit 2 (RPB2) genes. These genes also were compared for their utility in predicting phylogenetic relationships in this group of fungi. Maximum parsimony and Bayesian analyses conducted on separate and combined data sets indicate that ascospore morphology is extremely homoplastic and not useful for delimiting genera. Genera represented by more than one species were paraphyletic or polyphyletic in nearly all analyses; 17 species of Cercophora segregated into at least nine diVerent clades, while six species of Podos- pora occurred in Wve clades in the LSU tree. However, taxa with similar ascomal wall morphologies clustered in Wve well-supported clades suggesting that ascomal wall morphology is a better indicator of generic relationships in certain clades in the Sordariales. The RPB2 gene possessed over twice the number of parsimony-informative characters than either the LSU or -tubulin gene and conse- quently, provided the most support for the greatest number of clades.  2005 Elsevier Inc. All rights reserved.

Keywords: Ascomycota; Bayesian inference; -Tubulin; LSU; Morphological characters; RPB2; Phylogenetics; Sordariales; Systematics

1. Introduction decaying wood, leaf litter, and soil (Lundqvist, 1972). The Sordariales also was one of the most taxonomically The Sordariales is one of the most economically and diverse orders being comprised of 114 genera divided ecologically important groups within the ascomycetes in among 10 families (Eriksson and Hawksworth, 1998; that it contains species of Chaetomium, which are Eriksson et al., 2004), but recently has been reduced to responsible for the destruction of paper and fabrics, and ca. 35 genera within three families, the Chaetomiaceae, the “fruit Xies” of the fungal world (i.e., Neurospora Lasiosphaeriaceae, and Sordariaceae (Huhndorf et al., crassa, Podospora anserina, and Sordaria Wmicola). Taxa 2004). Since only one of these families (Sordariaceae) within the order occur worldwide as saprobes on dung, was shown to be monophyletic by Huhndorf et al. (2004), families within the Sordariales will not be further discussed. ¤ Corresponding author. Fax: +1 217 333 4949. The Sordariales is one of several orders in the Class E-mail address: [email protected] (A.N. Miller). Sordariomycetes (Eriksson et al., 2004). Taxa in the

Sordariomycetes (historically known as pyrenomyce- Ascomal wall morphology also has been suggested as tes) usually form minute fruiting bodies ( D ascomata) an alternative means of delimiting certain genera within containing hymenial layers commonly composed of this group (Lundqvist, 1972) (Fig. 2). All members of sterile hyphae intermixed among asci (with single wall Bombardia and Bombardioidea possess a similar ascomal layers) possessing ascospores (Alexopolous et al., 1996). wall referred to as a bombardioid wall, which contains a Few morphological characters exist with which to putatively stromatic ( D arising from vegetative hyphae) delimit taxa in the Sordariomycetes most likely due to gelatinized layer composed of interwoven hyphae their small stature and simple structure. Taxa within (Lundqvist, 1972) (Fig. 2C). Three other genera (Arnium, the Sordariomycetes have traditionally been distin- Cercophora, and Podospora) also contain species that guished based on characters of the ascomata and possess a similar gelatinized layer in their ascomal wall, ascospores, although centrum and ascus morphologies but since the wall is non-stromatic, it is termed pseudo- also have been used at higher taxonomic levels (Barr, bombardioid (Miller, 2003) (Figs. 2A and B). However, 1990; Luttrell, 1951; Parguey-Leduc and Janex-Favre, all of these species have been placed into diVerent genera 1981). Ascomata can have single- or multi-layered walls based primarily on diVerences in their ascospore mor- and may possess various types of outside covering such phologies (Fig. 1). Certain species of Cercophora and as tomentum, hairs, or setae. Although considerable Lasiosphaeria also have been placed into separate genera variation in ascomal wall morphology exists in the based on diVerences in their ascospore morphologies Sordariomycetes, its potential use in systematics has even though they possess similar three-layered ascomal seldom been recognized (Jensen, 1985). Several work- walls in which the outer layer is composed of hyphae ers, however, have noted similarities in ascomal wall that form a tomentum (Fig. 2D). Finally, certain species characters among taxa (von Arx et al., 1984; Barr, 1978; of Podospora possess ascomata with outer wall layers Carroll and Munk, 1964; Jensen, 1985; Lundqvist, that form swollen protruding cells or agglutinated hairs 1972). (Fig. 2E), and some of these species have been trans- Genera within the Sordariales have been delimited ferred into a separate genus, Schizothecium (Lundqvist, primarily on diVerences in their ascospore morphology 1972). These genera, which contain species that possess (Lundqvist, 1972) (Fig. 1). While ascospore morphology ascomata with obvious morphological diVerences in varies little within a genus, ascospores among genera in their ascomal walls, are the focus of this paper. Addi- the Sordariales range from a cylindrical, hyaline asco- tional genera in the Sordariales (e.g., Apiosordaria, Jugu- spore in Lasiosphaeria (Fig. 1A) to an ellipsoidal, brown lospora, and Triangularia), which contain species that ascospore in Sordaria (Fig. 1I). Intermixed between these possess ascomata with morphologically simple ascomal two extremes are many genera which possess two-celled walls, require further study and will be treated in future ascospores with cylindrical to ellipsoidal, brown cells studies. Our study is the Wrst to evaluate ascomal wall and diVerent degrees of cylindrical to triangular (often characters for their phylogenetic potential in delimiting basal), hyaline cells (Figs. 1B–G) (Lundqvist, 1972). Sev- certain genera in the Sordariales. eral earlier workers (Boedijn, 1962; Chenantais, 1919; Several nuclear and mitochondrial ribosomal and Lundqvist, 1972; Munk, 1953) hypothesized that asco- protein-coding genes have been employed for assessing spore evolution within this group may have occurred phylogenetic relationships of Wlamentous ascomycetes. along this continuum either through the gain or loss of a Nuclear ribosomal genes such as 18S small subunit hyaline cell, resulting in either Lasiosphaeria or Sordaria (SSU) and 28S LSU are commonly used due to their being the derived genus. ease in ampliWcation resulting from their high copy num-

Fig. 1. Ascospores of representative genera in the Sordariales. (A) Lasiosphaeria. (B) Cercophora. (C) Podospora. (D) Apiosordaria. (E) Triangularia. (F) ZopWella. (G) Jugulospora. (H) Bombardioidea. (I) Sordaria. Ascospore evolution has been hypothesized to have occurred through the loss (A ! I) or gain (I ! A) of a hyaline tail resulting in either Sordaria or Lasiosphaeria being the derived genus. Ascospores not to scale. 62 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Fig. 2. Ascomal walls occurring in members of the Wve wall clades (A–E); outer layer to the right. (A) Pseudo-bombardioid wall (Podospora Wmiseda). (B) Pseudo-bombardioid wall (Cercophora scortea). (C) Bombardioid wall (Bombardioidea anartia). (D) Three-layered wall with outer layer of hyphae forming tomentum (Lasiosphaeria ovina). (E) Wall with agglutinated hairs (Schizothecium vesticola). Ascomal walls not to scale. ber and the availability of numerous universal primers The primary purpose of this study was to test whether (Vilgalys and Hester, 1990; White et al., 1990). Most ascospore morphology is phylogenetically informative for studies utilize the Wrst 1100 bp of the 5Ј end of LSU, predicting generic relationships within the Sordariales which contains three variable domains. Nuclear protein- using a multi-gene approach. Multiple species, which pos- coding genes such as -tubulin and RPB2 are increas- sess the range of ascospore morphologies known to occur ingly being used in ascomycete phylogenetic studies in the order, were sampled from each of several genera. To incorporating multiple, unlinked genes. The -tubulin determine the phylogenetic potential of ascomal wall mor- gene contains a highly variable intron-rich 5Ј end and a phology for delimiting certain genera in the Sordariales, more conserved intron-poor 3Ј end, the latter of which species with similar ascomal walls in several genera also has been used to recover higher-level relationships in were included. Finally, a comparison of the LSU, -tubu- ascomycetes (Landvik et al., 2001). A paralogous copy lin, and RPB2 genes was made for their utility in resolving has been discovered in ascomycetes (Gold et al., 1991; phylogenetic relationships within this group of fungi. May et al., 1987; Panaccione and Hanau, 1990), but the duplication event is believed to have occurred after the divergence of the Sordariomycetes (Landvik et al., 2001). 2. Methods The RPB2 gene is the second largest subunit of the ribosomal polymerase II gene and contains 12 conserved 2.1. Taxon sampling sequence motifs interspersed among highly variable regions (Liu et al., 1999). Although a second copy has Taxa used in this study are listed in Table 1 along recently been found in plants (Oxelman and Bremer, with source information and origin for those specimens 2000), RPB2 is believed to occur as a single orthologous sequenced in this study. Multiple representatives were gene in fungi and is becoming increasingly popular in included from each of eight genera representing the studies of ascomycete phylogeny (Liu et al., 1999; Miller range in ascospore morphology in the Sordariales. Sev- and Huhndorf, 2004b; Reeb et al., 2004; Zhang and eral of these taxa also possess similar ascomal wall mor- Blackwell, 2002). phologies. The full data sets contained 95, 83, and 68 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 63

Table 1 Taxa used in this study Taxon Sourcea,b Originc GenBank Accession No.d LSU -Tubulin RPB2 Annulatascus triseptatus SMH2359 Costa Rica AY346257 AY780082 AY780148 Annulatascus triseptatus SMH4832 France AY780049 AY780083 — Annulatascus velatispora GenBank AF132320 — — Anthostomella sp. SMH3101 Puerto Rico AY780050e AY780084 — Apiosordaria backusii ATCC34568 Japan AY780051 AY780085 AY780149 Apiosordaria verruculosa F-152365 (A-12907) Spain AY346258 AY780086 AY780150 Barrina polyspora AWR9560A Texas AY346261 AY780087 — Bertia moriformis SMH4320 (a) Michigan AY695260 AY780088 AY780151 Bombardia bombarda AR1903 New Zealand AY780052e AY780089 AY780152f Bombardia bombarda SMH3391 Michigan AY346263 AY780090 AY780153f Bombardia bombarda SMH4821 France AY780053 AY780091 AY780154 Bombardioidea anartia HHB99-1 (a) Alaska AY346264 AY780092 AY780155 Botryosphaeria rhodina GenBank — — AF107802 Botryosphaeria ribis GenBank AY004336 — — Camarops amorpha SMH1450 Puerto Rico AY780054e AY780093 AY780156f Camarops petersii JM1655 (a) Indiana AY346265 AY780094 — Camarops tubulina SMH4614 (a) Denmark AY346266 AY780095 AY780157 Camarops ustulinoides SMH1988 (a) Puerto Rico AY346267 AY780096 — Capronia mansonii GenBank AY004338 — — Catabotrys deciduum SMH3436 Panama AY346268 AY780097 AY780158 Cercophora sp. SMH3200 Costa Rica AY780055e AY780098 AY780159f Cercophora areolata UAMH7495 Canada AY587936 AY600252 AY600275 Cercophora atropurpurea SMH2961 Puerto Rico AY780056e AY780099 — Cercophora atropurpurea SMH3073 Puerto Rico AY780057 AY780100 AY780160f Cercophora caudata SMH3298 North Carolina AY436407 AY780101 AY780161 Cercophora coprophila SMH3794 (a) Puerto Rico AY780058 AY780102 AY780162 Cercophora costaricensis SMH4021 (a) Costa Rica AY780059 AY780103 AY780163 Cercophora lanuginosa SMH3819 North Carolina AY436412 AY600262 AY600283 Cercophora macrocarpa SMH2000 Puerto Rico AY780060e — AY780164f Cercophora aV. mirabilis SMH4238 Costa Rica AY780061 AY780104 AY780165 Cercophora aV. mirabilis SMH4002 (a) Costa Rica AY346271 AY780105 — Cercophora newWeldiana SMH2622 Michigan AF064642 AF466019 AY780166f Cercophora newWeldiana SMH3303 North Carolina AY780062e AY780106 AY780167 Cercophora rugulosa SMH1518 Puerto Rico AY436414 AY600272 AY600294 Cercophora scortea GJS L556 Louisiana AY780063 AY780107 AY780168f Cercophora sordarioides UAMH9301 France AY780064 — — Cercophora sparsa JF00229 (a) France AY587937 AY600253 — Cercophora striata SMH3431 (a) Panama AY780065 AY780108 AY780169 Cercophora striata SMH4036 (a) Costa Rica AY780066 — — Cercophora sulphurella SMH2531 Illinois AY587938 AY600254 AY600276 Cercophora terricola ATCC200395 Japan AY780067 AY780109 AY780170 Chaetomium elatum GenBank — — AF107791 Chaetomium globosum SMH4214b Jamaica AY346272 AY780110 — Chaetomium microascoides F-153395 (A-12898) Spain AY346273 AY780111 AY780171 Chaetosphaerella phaeostroma SMH4585 (a) England AY346274 AY780112 AY780172 Chaetosphaeria innumera SMH2748 North Carolina AY017375 AF466018 — Chaetosphaeria ovoidea SMH2605 Michigan AF064641 AF466057 AY780173f Coniochaeta ligniaria SMH2569 Michigan AY346275 AY780113 — Coniochaetidium savoryi TRTC51980 Malawi AY346276 AY780114 AY780174 Copromyces sp. TRTC51747 (CBS386.78) Argentina AY346277 — — Daldinia concentrica GenBank U47828 — — Diaporthe phaseolorum FAU458 Mississippi AY346279 AY780115 AY780175 Diatrype disciformis GenBank U47829 — — Dothidea insculpta GenBank — — AF107800 Duradens sp. SMH1708 Puerto Rico AY780068e AY780116 — Eutypa sp. SMH3580 Panama AY346280 AY780117 AY780176 Gelasinospora tetrasperma ATCC96230 Canada AY346281 AY780118 AY780177f Hypocrea pallida GenBank — — AY015639 Hypocrea schweinitzii GenBank — — AY015636 (continued on next page) 64 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Table 1 (continued) Taxon Sourcea,b Originc GenBank Accession No.d LSU -Tubulin RPB2 Hypocrea virens GenBank — AY158203 — Hypomyces luteovirens GenBank AF160237 — — Hypomyces odoratus GenBank — Y12256 — Induratia sp. SMH1255 Puerto Rico AY780069 AY780119 — Jugulospora rotula ATCC38359 N/A AY346287 AY780120 AY780178 Lasiosphaeria glabrata TL4529 (a) Denmark AY436410 AY600255 AY600277 Lasiosphaeria hirsuta SMH1543 Wisconsin AY436417 AY780121 AY780179 Lasiosphaeria hispida SHM3336 North Carolina AY436419 AY780122 AY780180f Lasiosphaeria immersa SMH4104 Wisconsin AY436409 AY780123 AY780181 Lasiosphaeria ovina SMH1538 Illinois AF064643 AF466046 AY600287 Lasiosphaeria sorbina GJS L555 Louisiana AY436415 AY600273 AY600295 Lasiosphaeriella nitida SMH1664 Puerto Rico AY346289 AY780124 AY780182 Leptosphaeria maculans GenBank — AF257329 — Linocarpon appendiculatum ATCC90499 Brunei AY346291 AY780125 AY780183f Melanochaeta hemipsila SMH2125 Puerto Rico AY346292 AF466049 AY780184 Microascus trigonosporus GenBank — — AF107792 Nectriopsis violacea GenBank AF193242 — — Neurospora crassa GenBank AF286411 M13630 AF107789 Neurospora pannonica TRTC51327 Hungary AY780070 AY780126 AY780185f Nitschkia grevillei SMH4663 (a) Illinois AY346294 AY780127 — Ophioceras tenuisporum SMH1643 Puerto Rico AY346295 AY780128 — Pleospora herbarum GenBank — Y17077 — Podospora anserina GenBank — — AF107790 Podospora appendiculata CBS212.97 New Zealand AY780071 AY780129 AY780186f Podospora comata ATCC36713 Venezuela AY780072 — — Podospora decipiens CBS258.64 Wyoming AY780073 AY780130 AY780187 Podospora Wbrinocaudata TRTC48343 California AY780074e AY780131 AY780188 Podospora Wmbriata CBS144.54 N/A AY780075 AY780132 AY780189f Podospora Wmiseda CBS990.96 New Zealand AY346296 AY780133 AY780190f Poroconiochaeta discoidea SANK12878 Japan AY346297 AY780134 AY780191 Pseudohalonectria lignicola SMH2440 Costa Rica AY346299 AY780135 — Schizothecium curvisporum ATCC36709 Kenya AY346300 AY780136 AY780192 Schizothecium vesticola SMH3187 Indiana AY780076e —— Sinosphaeria bambusicola SMH1999 Puerto Rico AY780077e AY780137 AY780193 Sordaria humana ATCC22796 Oklahoma AY780078 — — Sordaria Wmicola SMH4106 (a) Wisconsin AY780079 AY780138 AY780194 Sordaria lappae SMH4107 (a) Wisconsin AY780080 AY780139 — Sordaria macrospora Buck s.n. Canada AY346301 AY780140 AY780195f Strattonia carbonaria ATCC34567 Japan AY346302 AY780141 AY780196f Striatosphaeria codinaeaphora SMH1524 Puerto Rico AF466088 — — Triangularia mangenotii ATCC38847 Japan AY346303 AY780142 — Triangularia tanzaniensis TRTC51981 Tanzania AY780081e AY780143 AY780197 Valsa ceratosperma AR3426 Austria AF408387 AY780144 AY780198 Valsonectria pulchella SMH1193 Puerto Rico AY346304 AY780145 AY780199f Xylaria hypoxylon GenBank U47841 — — ZopWella ebriosa CBS111.75 N/A AY346305 AY780146 AY780200 Zygopleurage zygospora SMH4219 Texas AY346306 AY780147 — a (a) D DNA extracted from ascomata; all others were extracted from cultures. b ATCC, American Type Culture Collection; CBS, Centraalbureau voor Schimmelcultures, Netherlands; TRTC, Royal Ontario Museum, Toronto, Canada; UAMH, University of Alberta Microfungus Collection and Herbarium; AR, Amy Rossman; AWR, A. W. Ramaley; Buck, Wil- liam Buck; FAU, Francis A. Uecker; GJS, Gary J. Samuels; HHB, Harold H. Burdsall; JF, Jacques Fournier; JM, Jack Murphy; SMH, Sabine M. Huhndorf; TL, Thomas Læssøe. c Origin not given for taxa obtained from GenBank. d Dashes indicate gene was not sequenced for taxon. e For these taxa, although 1100 bp were used in the analyses, 1300 bp were sequenced and deposited in GenBank. f For these taxa, although a 1200 bp region between conserved motifs 5 and 7 (Liu et al., 1999) was used in the analyses, an 1800 bp region between conserved motifs 3 and 7 was sequenced and deposited in GenBank. taxa for the LSU, -tubulin, and RPB2 genes, respec- in the combined analyses. Based on results from previ- tively, while reduced data sets sampled the same 58 taxa ous phylogenetic analyses (Huhndorf et al., 2004; Liu et al., for each of the three genes and were subsequently used 1999; Miller and Huhndorf, 2004a), two representatives of A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 65 loculoascomycetes were used as outgroups for the full 72 °C for 10min. Parameters for amplifying the protein- data sets, while a member of the Xylariales was used to coding genes were identical except annealing was con- root trees in the reduced data sets. All voucher speci- ducted at 50 °C for -tubulin and at 50 °C for 10 cycles mens are deposited in the Field Museum Mycology Her- followed by 20–30 cycles at 54–58 °C for RPB2. Ready- barium (F). To-Go PCR beads (Amersham–Pharmacia Biotech) were occasionally used to amplify diYcult taxa according 2.2. Morphological analyses to the manufacturer’s instructions. In rare cases of weak ampliWcation, a punch of the PCR product was taken All taxa in which voucher specimens were available from the gel, suspended in 50–150L double distilled (i.e., AR, AWR, GJS, HHB, JF, JM, SMH, and TL spec- sterile water, melted at 70 °C, and 1 L of this dilution imens) were used in morphological analyses (Table 1). was reampliWed using the thermocycling parameters Ascospore morphology was observed from squash above except the annealing temperature was increased 3– mounts of ascomata made in water, while ascomal wall 5 °C. After veriWcation on an ethidium bromide-stained morphology was determined from sections made at ca. 1% TBE agarose gel, PCR products were gel-puriWed on 30 m following the techniques of Miller (2003). Images a 1% TALE agarose gel using GELase Agarose Gel- were captured using diVerential interference (DIC) Digesting Preparation (Epicentre Technologies). A Big- microscopy from a Dage DC-330 video system mounted Dye Terminator Cycle Sequencing Kit (ABI PRISM, on a Zeiss Axioskop and processed using Adobe Photo- Perkin–Elmer Biosystems) was used to sequence both shop 3.0 and 5.5 (Adobe Systems). strands using a combination of the following primers: LSU D LROR, LRFF1, LRAM1, LR3, LR3R, LR5, and 2.3. DNA extraction, ampliWcation, and sequencing LR6 (Huhndorf et al., 2004; Rehner and Samuels, 1995; Vilgalys and Hester, 1990); -tubulin D BT1819R, Bt1a, A DNeasy Mini Plant extraction kit (Qiagen) was BT1283, BT1283R, BTAM1f, BTAM1R, and BT2916 used for extracting DNA from either dried ascomata or (Glass and Donaldson, 1995; Table 2); and cultures following the manufacturer’s protocols except RPB2 D fRPB2-5f, RPB2AM-6R, RPB2AM-1f, tissues were ground in 100 L AP1 buVer instead of liq- RPB2AM-1R, RPB2AM-1bf, RPB2AM-1bR, and uid nitrogen. The relative quantity of total genomic DNA RPB2AM-7R (Table 2). Sequences were generated on an was observed on a 1% TBE agarose gel stained with ethi- Applied Biosystems 3100 automated DNA sequencer. dium bromide. Gene fragments were PCR-ampliWed on Each sequence fragment was subjected to a blast search either a MJ Research PTC 200 or PTC 220 Dyad thermo to verify its identity. Sequences were assembled and cycler using the following oligonucleotide primers: aligned with Sequencher 4.1 (Gene Codes), optimized by LSU D LROR–LR7 (Rehner and Samuels, 1995; Vilgalys eye, and manually corrected when necessary. and Hester, 1990), -tubulin D BT1819R–BT2916 (Table 2), and RPB2D fRPB2-5f–RPB2AM-7R (Liu et al., 1999; 2.4. Phylogenetic analyses Table 2). The LSU was ampliWed using the following thermocycling parameters: initial denaturation at 94 °C 2.4.1. Saturation for 2 min followed by 35–40 cycles of 94 °C for 30 s, 47 °C A considerable number of changes occur in the third for 15 s, and 72 °C for 1 min with a Wnal extension step of codon positions compared to the Wrst and second

Table 2 Primers developed in this study for PCR ampliWcation and sequencing Name Primer sequence Positiona BT1819Rb 5Ј-TTC CGT CCC GAC AAC TTC GT-3Ј 1131–1150 BT1283c 5Ј-CGC GGG AAG GGC ACC ATG TTG-3Ј 1643–1663 BT1283Rb 5Ј-CAA CAT GGT GCC CTT CCC GCG-3Ј 1643–1663 BT2916b 5Ј-CTC AGC CTC AGT GAA CTC CAT-3Ј 2151–2171 BTAM1f 5Ј-GTT CGA CCC CAA GAA CAT GAT GGC YGC-3Ј 1757–1783 BTAM1R 5Ј-GCA GCC ATC ATG TTC TTG G-3Ј 1765–1783 RPB2AM-6R 5Ј-TTG ACC AGA CCR CAA GCC TG-3Ј 985–1004 RPB2AM-1f 5Ј-GAG TTC AAG ATY TTC TCK GAT GC-3Ј 1261–1283 RPB2AM-1R 5Ј-GCA TCM GAG AAR ATC TTG AAC TC-3Ј 1261–1283 RPB2AM-1bf 5Ј-CCA AGG TBT TYG TSA ACG G-3Ј 1127–1145 RPB2AM-1bR 5Ј-GGY CTC ATR ACR CGR CCR GC-3Ј 1282–1301 RPB2AM-7R 5Ј-GAA TRT TGG CCA TGG TRT CCA T-3Ј 1783–1804 Letters follow standard IUPAC–IUBMB ambiguity codes. a Relative to Neurospora crassa as sequenced in Orbach et al. (1986) for -tubulin (M13630) and in Liu et al. (1999) for RPB2 (AF107789). b Developed by Valérie Reeb in Lutzoni lab, Biology Dept., Duke University (http://www.lutzonilab.net/pages/primer.shtml). c Developed by Fernando Fernández in Huhndorf lab, Botany Dept., The Field Museum of Natural History. 66 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Table 3 Comparison of data sets and trees in phylogenetic analyses Data sets Genes Combined LSU -Tubulin RPB2 Full Reduced Full Reduced Full Reduced No. of total sitesa 1084 1039 939 924 1200 1197 3153 No. of ambiguous sites 222 126 0 0 222 222 348 No. of constant sites 524 599 568 591 372 407 1596 No. of polymorphic sites 338 314 371 333 606 568 1214 No. of parsimony-informative sitesb 251 228 41, 16, 254 (311) 27, 8, 231 (266) 163, 84, 312 (559) 153, 75, 310 (538) 1032 Percent of total sites that are 23 22 33 29 47 45 33 parsimony-informative No. of MP trees 2 1 2 1 1 1 1 Length of MP trees 3070.46 2070.52 4532.81 3164.12 12751.79 10242.09 15660.21 Consistency index 0.370 0.402 0.207 0.248 0.174 0.195 0.231 Retention index 0.632 0.583 0.509 0.471 0.449 0.467 0.475 Rescaled consistency index 0.234 0.235 0.105 0.117 0.078 0.091 0.110 No. of clades with 770% bootstrap support 39 22 30 19 32 29 35 No. of clades with 795% Bayesian post. prob. 35 26 27 20 39 40 40 a Excluding sites in 5Ј and 3Ј ends and introns. b Divided into Wrst, second, and third codon positions for -tubulin and RPB2; total shown in parentheses. positions in the two protein-coding genes, especially - ping using PAUP* 4.0b10 (SwoVord, 2002). tubulin (Table 3), suggesting these sites may be saturated Unambiguously aligned characters in the LSU data sets and, thus, represent noise rather than phylogenetic sig- and each of the three codon positions in the -tubulin nal. Therefore, analyses were conducted on the full - and RPB2 data sets were subjected to a symmetric tubulin and RPB2 data sets to determine if the Wrst, stepmatrix generated using STMatrix ver. 2.2 (François second, and third codon positions were saturated by Lutzoni and Stefan Zoller, Biology Department, Duke constructing scatter plots which compare time of University), which calculates the costs for changes sequence divergence to pairwise transition and pairwise among character states based on the negative natural transversion divergences (Hackett, 1996). Uncorrected logarithm of the percentages of reciprocal changes pairwise sequence divergence (uncorrected “p”) was between any two character states. The phylogenetic sig- used as an approximation of divergence time. Transi- nal from 11 of the 15 ambiguous regions in the full LSU tions and transversions at each of the codon positions data set and Wve of the seven regions in the reduced LSU were determined to be saturated if the scatter of points data set was recovered using INAASE (Lutzoni et al., appeared to level oV as sequence divergence increased. 2000) and analyzed in the MP analyses. The remaining LSU ambiguous regions and the two RPB2 ambiguous 2.4.2. Maximum parsimony and Bayesian analyses regions were excluded because their recoded characters Maximum parsimony (MP) analyses were performed contained more than 32 character states, which is not on each of the three full data sets and on the equal-sized allowed in PAUP*. Branch support for all MP analyses (58 taxa) reduced data sets to assess the amount of was estimated by performing 1000 bootstrap replicates incongruence among data partitions (see below) and to (Felsenstein, 1985), each consisting of 100 random addi- compare the relative utility of the three genes in resolv- tion heuristic searches and TBR branch-swapping. ing relationships. Portions of the 5Ј and 3Ј ends of each MODELTEST 3.06 (Posada and Crandall, 1998) was data set were excluded from all analyses due to missing used to determine the best-Wt model of evolution for data in most taxa. Fifteen and seven ambiguously each data set. Bayesian analyses employing a Markov aligned regions were delimited in the full and reduced chain Monte Carlo (MCMC) method were performed LSU data sets, respectively, and these regions along with using MrBayes 3.0b4 (Huelsenbeck and Ronquist, 2001) three introns were excluded from all analyses. Single as an additional means of assessing branch support. The introns in the -tubulin and RPB2 data sets also were best-Wt model of evolution was implemented for each excluded from all analyses. Several taxa in the RPB2 data set in the separate analyses and for each partition data sets contained amino acid indels in the highly vari- (i.e., separate models for LSU and for each of the three able region between conserved motifs 6 and 7 (Liu et al., codon positions in -tubulin and RPB2) in the combined 1999). These regions were so variable that even amino analyses. Constant characters were included and four acids could not be unambiguously aligned, so these MCMC chains were ran simultaneously for 5,000,000 regions were excluded from all analyses. Unequally generations with trees sampled every 100th generation weighted MP analyses were performed with 1000 ran- resulting in 50,000 total trees. The MCMC chains always dom addition heuristic searches and TBR branch-swap- achieved stationarity after the Wrst 20,000–150,000 gen- A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 67 erations ( D 200–1500 trees), so the Wrst 10,000 trees, which extended well beyond the burn-in phase in each analysis, were discarded. Posterior probabilities were determined from a consensus tree generated from the remaining 40,000 trees. This analysis was repeated three times starting from diVerent random trees to ensure trees from the same tree space were being sampled during each analysis.

2.4.3. Combinability The validity of the incongruence length diVerence (ILD) test for determining whether multiple data sets should be combined has recently been questioned (Barker and Lutzoni, 2002; Yoder et al., 2001) and, thus, other methods should be explored. One method of assessing combinability of data sets, and the one adopted in this study, is by simply comparing highly supported clades among trees generated from diVerent data sets to detect conXict (de Queiroz, 1993; Mason- Gamer and Kellogg, 1996). High support typically refers to bootstrap support values 770% and Bayesian posterior probabilities 795% (Alfaro et al., 2003). If no con- Xict exists between the highly supported clades in trees generated from these diVerent data sets, this suggests the genes share similar phylogenetic histories and phylogenetic resolution and support could ultimately be increased by combining the data sets.

3. Results

3.1. Phylogenetic analyses

3.1.1. Saturation Except for third position transitions in the RPB2 gene, no evidence of saturation was detected in any of the codon positions since scatter plots clearly show an increase when pairwise sequence divergence is plotted against pairwise transition/transversion divergence (Fig. 3). Only third positions were plotted for the -tubulin gene since very few changes occur in the Wrst and second codon positions (Table 3). The scatter plot of third position transitions in the RPB2 gene appears to be leveling oV slightly suggesting a low level of saturation may be Fig. 3. Saturation plots relating uncorrected pairwise sequence divergence to pairwise transition/transversion divergence. Only third posi- occurring at these sites (Fig. 3). As expected, the scatter tion changes are shown for -tubulin (A), whereas Wrst, second, and of points in this area (points at the far right of the graph) third position transitions (B), and transversions (C) are shown for primarily represent pairwise comparisons between the RPB2. Sordariales ingroup taxa and the more distant loculoascomycetes outgroups. Therefore, additional MP analyses 1.86). A single most-parsimonious tree with an identical were conducted on the RPB2 full data set in which third topology to that in Fig. 6 was estimated when transitions position transitions were arbitrarily down-weighted by a were down-weighted by 2. However, phylogenetic reso- factor of 2, 10, and 100 relative to transversions using a lution was substantially decreased in analyses in which stepmatrix. Third position transitions and transversions transitions were down-weighted by 10 and 100 in that were weighted approximately equally in the original rate outgroup taxa (Coniochaetales and Chaetosphaeriales) substitution stepmatrix (A M C D 1.84, A M G D 1.77, occurred within the ingroup (Sordariales) (data not A M T D 1.86, C M G D 1.82, C M T D 1.63, and G M T D shown). This suggests that while some third position 68 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 transitions are becoming saturated, a majority of these dariales in all analyses (Figs. 4–7). For example, 17 spe- sites still possess phylogenetic signal which is necessary cies of Cercophora segregate into at least nine diVerent to accurately estimate phylogenies. clades, while six species of Podospora occur in Wve clades in the LSU tree (Fig. 4). Multiple species were sampled 3.1.2. Maximum parsimony and Bayesian analyses from six to nine genera (Apiosordaria, Cercophora, Chae- The best-Wt maximum-likelihood model of evolution tomium, Lasiosphaeria, Neurospora, Podospora, Schizo- for the LSU data sets was the TIMef model (Rodríguez thecium, Sordaria, and Triangularia) in the full data sets et al., 1990). The best-Wt model for the -tubulin data and all genera occur as paraphyletic or polyphyletic in at sets and the full RPB2 data set was the GTR model least one of the most-parsimonious trees (Figs. 4–6). (Rodríguez et al., 1990), while the TrN model (Tamura Combined analyses of the reduced data sets corroborate and Nei, 1993) was selected as the best-Wt model for the these results for four (Cercophora, Lasiosphaeria, Neu- reduced RPB2 data set. A proportion of sites were rospora, and Podospora) of the six genera represented by invariable while the remaining sites were subjected to a multiple species (Fig. 7). While morphology suggests gamma distribution shape parameter in all models. that ascospore evolution in the Sordariales may have Applying separate models to each of the three codon occurred directionally along a continuum (Fig. 1), our positions for the -tubulin and RPB2 data sets had little molecular data do not support this in that species with eVect on the Bayesian posterior probabilities in the com- vastly diVerent ascospore morphologies occur in several bined analyses (data not shown). well-supported clades throughout the Sordariales (Figs. Maximum parsimony analyses of the LSU and - 4–7). Three well-supported clades exist which contain tubulin full data sets each generated two trees. The LSU species of Cercophora that cluster with species of Apios- trees diVered only slightly in relationships among species ordaria or Triangularia (Figs. 4 and 5), taxa which pos- of Sordaria, whereas the -tubulin trees diVered in rela- sess ascospore morphologies that do not occur along the tionships among species of Neurospora. One of the trees putative evolutionary transition. The most extreme from each data set is shown in Figs. 4 and 5. A single example is found in the well-supported clade that con- most-parsimonious tree was generated in analyses of the tains Bombardia and Bombardioidea (Figs. 4–7). While RPB2 full data set (Fig. 6). Data set and tree statistics Bombardia possesses Cercophora-like ascospores at one are listed in Table 3. Overall relationships among the end of the continuum (Fig. 1B), Bombardioidea possesses major ordinal and familial lineages diVered between the ellipsoidal, brown ascospores at the opposite end of the most-parsimonious trees generated from the full data continuum (Fig. 1H). sets, although most of these relationships were not supported (Figs. 4–6). 3.3. Ascomal wall morphology

3.1.3. Combinability Although many of the clades within the Sordariales are Separate analyses of the three reduced data sets pro- unsupported, Wve well-supported clades, which contain duced single most-parsimonious trees (data not shown). species with similar ascomal wall morphologies, occur in Data set and tree statistics are listed in Table 3 for the all trees (Figs. 4–7). Species of Cercophora and Podospora, reduced data sets. Although minor diVerences occurred which possess similar pseudo-bombardioid walls (Figs. among most-parsimonious trees in the reduced data sets 2A and B), occur in two of these clades (wall clades A and (data not shown), only one instance of highly supported B), while species of Bombardia and Bombardioidea with conXict existed between phylogenies. In the LSU tree, the similar bombardioid walls (Fig. 2C) occur in a third clade Coronophorales were a sister clade to the Hypocreales (wall clade C). A fourth clade (wall clade D) is repre- with 77% bootstrap support, whereas this order was sented by species of Cercophora and Lasiosphaeria which placed within the Boliniales with 79% bootstrap support possess a similar three-layered ascomal wall in which the in the -tubulin tree (data not shown). This conXict was outer wall layer is composed of hyphae that form a not supported by Bayesian posterior probabilities and is tomentum (Fig. 2D), while the Wfth clade (wall clade E) is most likely due to poor taxon sampling in these orders. represented by species of Podospora/Schizothecium which Since little evidence exists against combining these data possess an outer ascomal wall layer that forms swollen sets, they were analyzed simultaneously in a combined protruding cells or agglutinated hairs (Fig. 2E). analysis, which produced a single most-parsimonious tree (Fig. 7). Combined data statistics are given in Table 3. 4. Discussion 3.2. Ascospore morphology 4.1. Ascospore morphology Ascospore morphology is shown to be extremely homoplastic in that multiple species in the same genus The Sordariales has recently been redeWned to include occur in diVerent clades scattered throughout the Sor- genera which possess ascospores that share a similar A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 69

Fig. 4. One of two most-parsimonious trees based on the LSU data set of 95 taxa. Numbers above or below the branches indicate bootstrap support based on 1000 replicates. Thickened branches represent signiWcant posterior probabilities (795%) generated from Bayesian analyses. Shaded boxes indicate the Wve well-supported ascomal wall clades discussed in the text. ClassiWcation following Huhndorf et al. (2004) is shown along the right. developmental pattern (i.e., cylindrical, hyaline ascosp- (Figs. 5 and 6). These genera possess ascospores which ores that develop apical, brown heads and basal, hyaline form a continuum from a cylindrical, hyaline ascospore tails) (Huhndorf et al., 2004). Seventeen genera were in Lasiosphaeria to an ellipsoidal, brown ascospore in included in the LSU data set which possess ascospores Sordaria with numerous genera which possess ascosp- that vary along these developmental lines and all Wnd ores with brown heads and hyaline tails intermixed their placement in the monophyletic Sordariales (Fig. 4). between (Fig. 1). This hypothesized trend in ascospore The -tubulin and RPB2 data sets, which included 16 evolution is, however, not supported by our molecular and 15 genera, respectively, corroborate these results data (Figs. 4–7). Since diVerent taxa occur as the basal 70 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Fig. 5. One of two most-parsimonious trees based on the -tubulin data set of 83 taxa. Support values, shading, and classiWcation as in Fig. 4. group of the order in diVerent trees, any trend, if one relationships within the group. When more than one spe- even exists, in ascospore evolution is presently unclear. cies from the same genus was included, most did not Genera within the Sordariales have been delimited resolve along current generic lines, but were scattered primarily on diVerences in their ascospore morphology throughout several clades (Figs. 4–7). Dettman et al. (Lundqvist, 1972). While our data conWrm the Wndings (2001) also found ascospore morphology to be a poor of Huhndorf et al. (2004) that ascospore morphology is a predictor of phylogenetic relationships in Neurospora good indicator of whether a taxon belongs in the Sordar- and Gelasinospora. Analyses of four nuclear genes iales, these data indicate that ascospore morphology is revealed that the two genera conventionally distin- extremely homoplastic and a poor predictor of generic guished by diVerences in ascospore ornamentation did A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 71

Fig. 6. Single most-parsimonious tree based on the RPB2 data set of 68 taxa. Support values, shading, and classiWcation as in Fig. 4. not represent two distinct monophyletic groups but on ascospore shape and ornamentation and, as in our instead were a polyphyletic group of closely related, study, LSU sequences suggest that ascospore morphol- morphologically similar taxa. The two species of Neuros- ogy is a poor predictor of generic relationships (Kurtz- pora included in our study never occurred as a mono- man and Robnett, 1994). Additional studies are needed phyletic clade in any analysis (Figs. 4–7). to determine whether ascospore morphology is phyloge- The Sordariales is not the only group where asco- netically informative for delimiting genera in other spore morphology has been used for delimiting genera. groups of ascomycetes. For example, generic delimitation within the Amphisph- aeriaceae (Xylariales) has been based on ascospore pig- 4.2. Ascomal wall morphology mentation, septation, and shape and, in some cases, ornamentation (Barr, 1994; Kang et al., 1999). Genera of Although ascospore morphology is a poor predictor ascosporogenous yeasts also have been based primarily of generic relationships in the Sordariales, ascomal wall 72 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Fig. 7. Single most-parsimonious tree based on the combined data set of 58 taxa. Support values, shading, and classiWcation as in Fig. 4. morphology may serve as an alternative means of delim- Additional taxa must be included and further examina- iting certain genera. Five highly supported clades (wall tion of their ascomal walls must be completed before clades A–E) are found in all trees (Figs. 4–7) and four of their homologous characters can be adequately discussed. these clades (wall clades A–D) contain taxa with diVer- Three well-supported clades (wall clades A–C) con- ent ascospore morphologies but similar ascomal wall tain taxa possessing an ascomal wall with a gelatinized morphologies. Because taxon sampling within each of layer (Figs. 2A–C). Wall clades A and B contain species these Wve wall clades is nearly complete, it is unlikely of Cercophora and Podospora which possess a pseudo- that the addition of more Sordarialean taxa will alter bombardioid wall (Figs. 2A and B), while wall clade C current hypotheses of relationships in these clades. contains species of Bombardia and Bombardioidea which Therefore, discussions of their homologous characters have a bombardioid wall (Miller, 2003) (Fig. 2C). are appropriate. However, taxon sampling for other Ascospores in Cercophora are initially cylindrical but well-supported clades throughout the Sordariales is eventually develop a swollen head and long tail (Fig. presently incomplete and relationships may change with 1B), whereas those in Podospora are initially clavate the addition of more taxa. In addition, taxa in these before developing a swollen head and short tail (Fig. clades possess relatively simple 2- to 3-layered ascomal 1C). The pseudo-bombardioid wall morphology is walls, which possess no obvious morphological charac- slightly homoplasious in the Sordariales in that it ters with which to distinguish them at the present time. appears to have arisen independently in two distantly A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 73 related groups, although relationships among these 3- to 4-layered ascomal walls and ellipsoidal, brown groups are unsupported at this time (Figs. 4–7). How- ascospores (Lundqvist, 1972). Representatives of these ever, taxa in wall clade B can be diVerentiated from genera always formed a highly supported clade in our those in wall clade A in possessing short, brown, hyaline- study (Figs. 4–7). However, since these ascomal wall and tipped setae. Species of Bombardia and Bombardioidea ascospore morphologies occur in several other taxa are morphologically similar in all characters except their throughout the Sordariales (Lundqvist, 1972), it is pres- ascospores, which occur near the extremes of the mor- ently unclear which characters are phylogenetically phological ascospore continuum. Ascospores in Bombar- informative for delimiting this clade. dia are identical to those in Cercophora (Fig. 1B), One may argue that only characters which are puta- whereas those in Bombardioidea are ellipsoidal and tively homologous ( D synapomorphies) should be used brown (Fig. 1H). However, these species form a highly for delimiting taxa. However, this argument ignores the supported clade (wall clade C) united by a bombardioid fact that there are diVerent degrees of homoplasy. Char- ascomal wall (Fig. 2C), which occurs as a homologous acters can range from slightly homoplasious (arising in character in the Sordariales (Figs. 4–7). only two distantly related groups) to extremely homo- A fourth wall clade (wall clade D) contains species of plasious (arising in several groups throughout a tree) Cercophora and Lasiosphaeria. While both genera pos- with certain levels of the former still contributing some sess similar ascospores that are cylindrical and hyaline, amount of phylogenetic structure to the data set. At those in Cercophora eventually swell at one end and turn what level homoplasy stops becoming partially informa- brown (Fig. 1B). However, species in this clade possess a tive and becomes merely noise is presently unclear. similar three-layered ascomal wall in which the outer Although ascomal wall morphology is slightly homo- layer is composed of hyphae that form a tomentum (Fig. plasious in some groups (i.e., tomentum wall, pseudo- 2D). This ascomal wall morphology is slightly homo- bombardioid wall), ascospore morphology is extremely plastic in that three additional taxa (C. coprophila, C. homoplasious throughout this group. Thus, while asco- sparsa, and C. sulphurella), which also possess morpho- spore morphology cannot be used for delimiting genera, logically similar walls, are not found within this well- ascomal wall morphology alone or in combination with supported tomentum clade (Figs. 4–7). Cercophora other characters is still useful at some level for distin- coprophila occurs well outside this clade, while C. sparsa guishing taxa. and C. sulphurella occur as unsupported sister taxa to this clade. However, the wall in C. coprophila has been 4.3. Comparison of genes interpreted by some to be slightly areolate (Lundqvist, 1972) and this distinction may separate it from species in All three genes were compared using the same 58 wall clade D. taxa in the reduced data sets, which contain approxi- The Wfth wall clade (wall clade E) includes species of mately the same number of total sites (i.e., 924– Podospora/Schizothecium which possess an outer asco- 1197 bp) (Table 3). However, RPB2 contains over mal wall layer that forms swollen protruding cells or twice the number of parsimony-informative sites (538) agglutinated hairs (Fig. 2E). Other species of Podospora as LSU (228) or -tubulin (266) (Table 3) resulting in included in this study possess glabrous ascomata or are longer branch lengths and increased support for covered with short to long, Xexuous hairs or stiV setae clades. RPB2 contains more clades with signiWcant (Lundqvist, 1972). Schizothecium can be further distin- bootstrap support (29) and Bayesian posterior proba- guished from Podospora by the absence of typical inte- bilities (40) than either LSU (22, 26) or -tubulin (19, rascal paraphyses and ascospores which become septate 20). LSU possesses numerous indels in its Wrst three at a very early stage in their development (Lundqvist, domains thereby reducing the number of parsimony- 1972), but some believe these characters do not warrant informative sites after ambiguous regions are generic distinction (Bell and Mahoney, 1995). Although removed. Most of the phylogenetic signal from - wall clade E may presently be delimited by a combina- tubulin comes from third position changes (87%), tion of ascomal wall, centrum, and ascospore characters, whereas third positions account for only a little over additional species putatively belonging in Schizothecium half (57%) of the signal in the RPB2 gene. Despite the should be included in future analyses to test the signiW- extreme bias towards changes in third position sites in cance of these characters in this clade. -tubulin, these sites showed no evidence of saturation While studying the surface morphology of outer asco- for either transitions or transversions (Fig. 3). mal walls, Jensen (1985) found similarities among repre- Although changes are more evenly distributed sentative species from several orders and families of throughout Wrst, second, and third codon positions in Sordariomycetes. He also noted that three genera in the RPB2, a low level of saturation was detected in third Sordariaceae (i.e., Gelasinospora, Neurospora, and Sorda- position transitions. Saturation of third position tran- ria) possessed virtually identical outer ascomal wall sitions and transversions in RPB2 also was found by layers. These genera also possess similar membraneous, Reeb et al. (2004) in their study of euascomycetes. 74 A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75

Future studies employing RPB2 for estimating fungal as two anonymous reviewers for their comments which phylogenies should determine the level of saturation improved this paper. Sequences were generated in the at third positions and its eVect on the resolution of the Pritzker Laboratory for Molecular Systematics and Evo- resulting phylogeny. lution at FMNH. This work represents a portion of a the- sis in partial fulWllment of the requirements for the doctoral degree at the Graduate College of UIC. 5. Conclusion

This study contributes to the understanding of the References evolution of morphological characters in ascomycetes. Based on results from MP and Bayesian analyses of Alexopolous, C.J., Mims, C.W., Blackwell, M., 1996. Introductory three genes, ascospores with similar morphologies Mycology, fourth ed. John Wiley, New York. appear to have evolved independently numerous times Alfaro, M.E., Zoller, S., Lutzoni, F., 2003. Bayes or bootstrap. A simu- throughout the Sordariales, whereas ascomal walls lation study comparing the performance of Bayesian Markov chain Monte Carlo sampling and bootstrapping in assessing phylogenetic appear to be less homoplasious in this group. Relation- conWdence. Mol. Biol. Evol. 20, 255–266. ships among taxa outside the wall clades are mostly von Arx, J.A., Dreyfuss, M., Müller, E., 1984. A revaluation of Chaeto- unresolved and most relationships among well-sup- mium and the Chaetomiaceae. Persoonia 12, 169–179. ported clades are currently unsupported making it diY- Barker, F.K., Lutzoni, F.M., 2002. The utility of the incongruence V cult to draw conclusions regarding the evolution of length di erence test. Syst. Biol. 51, 625–637. Barr, M.E., 1978. The Diaporthales in North America. Mycol. Mem- ascomal wall and ascospore characters. Increased taxon oirs 7, 1–232. sampling will improve resolution in many clades, while Barr, M.E., 1990. Prodromus to nonlichenized, pyrenomycetous mem- support for relationships will be increased through the bers of Class Hymenoascomycetes. Mycotaxon 39, 43–184. incorporation of additional genes. While genera cannot Barr, M.E., 1994. Notes on Amphisphaeriaceae and related families. be recognized based on ascospore morphology, they Mycotaxon 51, 191–224. Bell, A., Mahoney, D.P., 1995. Coprophilous fungi in New Zealand. I. may be delimited in certain clades based on ascomal wall Podospora species with swollen agglutinated perithecial hairs. Myc- morphology. However, we are not advocating simply ologia 87, 375–396. replacing one form of a one-character taxonomy based Boedijn, K.B., 1962. The Sordariaceae of Indonesia. Persoonia 2, 305– on ascospores with another based on ascomal walls; 320. redeWned genera would need to be delimited by a unique Carroll, G.C., Munk, A., 1964. Studies on lignicolous Sordariaceae. Mycologia 56, 77–98. combination of characters primarily involving, but not Chenantais, J.-E., 1919. Études sur les Pyrenomycètes. Bull. Soc. Mycol. limited to, those found in ascomal walls. Fr. 35, 46–98. de Queiroz, A., 1993. For consensus (sometimes). Syst. Biol. 42, 368–372. Dettman, J.R., Harbinski, F.M., Taylor, J.W., 2001. Ascospore mor- Acknowledgments phology is a poor predictor of the phylogenetic relationships of Neurospora and Gelasinospora. Fung. Genet. Biol. 34, 49–61. Eriksson, O.E., Hawksworth, D.L., 1998. Outline of the ascomycetes— This work was supported in part by a NSF DDIG 1998. Systema Ascomycetum 16, 83–296. Grant (Doctoral Dissertation Improvement Grant, DEB- Eriksson, O.E., Baral, H.-O., Currah, R.S., Hansen, K., Kurtzman, C.P., 0105077) to ANM through the University of Illinois at Rambold, G., Laessøe, T. (Eds.), 2004. Outline of Ascomycota— Chicago (UIC) and in part by NSF PEET Grants (Part- 2004. Myconet, vol. 10, pp. 1–99. Felsenstein, J., 1985. ConWdence intervals on phylogenies: An nerships for Enhancing Expertise in Taxonomy, DEB- approach using the bootstrap. Evolution 39, 783–791. 9521926 and DEB-0118695) to SMH through the Field Glass, N.L., Donaldson, G.C., 1995. Development of primer sets Museum of Natural History (FMNH). ANM also was designed for use with the PCR to amplify conserved genes from supported during this study by a Lester Armour Gradu- Wlamentous ascomycetes. Appl. Environ. Microbiol. 61, 1323–1330. ate Fellowship from FMNH. Fieldwork for ANM was Gold, S.E., Casale, W.L., Keen, N.T., 1991. Characterization of 2 beta- tubulin genes from Geotrichum candidum. Mol. Gen. Genet. 230 (1- supported in part by an ASPT (American Society of Plant 2), 104–112. Taxonomists) Graduate Student Research Grant and an Hackett, S.J., 1996. Molecular phylogenetics and biogeography of tan- UIC Provost Award. Fieldwork for SMH was supported agers in the genus Ramphocelus (Aves). Mol. Phylogenet. Evol. 5, in part by the National Research Council Resident 368–382. Research Associate Post-doctoral Program in coopera- Huelsenbeck, J.P., Ronquist, F.R., 2001. MrBayes: Bayesian inference of phylogenetic trees. Bioinformatics 17, 754–755. tion with the USDA Forest Service, Madison, WI. The Huhndorf, S.M., Miller, A.N., Fernández, F.A., 2004. Molecular sys- authors are most grateful to G. Bills, W. Buck, H. Burd- tematics of the Sordariales: The order and the family Lasiosphaeri- sall, M. Calduch, F. Fernández, J. Fournier, J. Krug, T. aceae redeWned. Mycologia 96 (2), 368–387. Læssøe, A. Rossman, G. Samuels, and A. Stchigel for pro- Jensen, J.D., 1985. Peridial anatomy and pyrenomycete taxonomy. viding specimens or cultures. V. Reeb and F. Fernández Mycologia 77, 688–701. Kang, J.C., Hyde, K.D., Kong, R.Y.C., 1999. Studies on Amphisphaeri- graciously allowed us to use their unpublished primers. ales: The Amphisphaeriaceae (sensu stricto). Mycol. Res. 103, 53– We also wish to thank A. Mitchell and G. Mueller as well 64. A.N. Miller, S.M. Huhndorf / Molecular Phylogenetics and Evolution 35 (2005) 60–75 75

Kurtzman, C.P., Robnett, C.J., 1994. Orders and families of ascospor- Oxelman, B., Bremer, B., 2000. Discovery of paralogous nuclear gene ogenous yeasts and yeast-like taxa compared from ribosomal RNA sequences coding for the second-largest subunit of RNA polymer- sequence similarity. In: Hawksworth, D.L. (Ed.), Ascomycete Sys- ase II (RPB2) and their phylogenetic utility in Gentianales of the tematics: Problems and Perspectives in the Nineties. Plenum, New Asterids. Mol. Biol. Evol. 17, 1131–1145. York, pp. 249–258. Panaccione, D.G., Hanau, R.M., 1990. Characterization of two diver- Landvik, S., Eriksson, O.E., Berbee, M.L., 2001. Neolecta—A fungal gent beta-tubulin genes from Colletotrichum graminicola. Gene 86 dinosaur. Evidence from beta-tubulin amino acid sequences. Myco- (2), 163–170. logia 93, 1151–1163. Parguey-Leduc, A., Janex-Favre, M.C., 1981. The ascocarps of ascohy- Liu, Y.J., Whelen, S., Hall, B.D., 1999. Phylogenetic relationships menial Pyrenomycetes. In: Reynolds, D. (Ed.), Ascomycete System- among ascomycetes: Evidence from an RNA Polymerase II Sub- atics. Springer-Verlag, New York, pp. 102–123. unit. Mol. Biol. Evol. 16, 1799–1808. Posada, D., Crandall, K.A., 1998. Modeltest: Testing the model of Lundqvist, N., 1972. Nordic Sordariaceae s. lat. Symb. Bot. Ups. 20, 1– DNA substitution. Bioinformatics 49, 817–818. 374. Reeb, V., Lutzoni, F., Roux, C., 2004. Contribution of RPB2 to mul- Luttrell, E.S., 1951. Taxonomy of the pyrenomycetes. University of tilocus phylogenetic studies of the euascomycetes (Pezizomycotina, Missouri Studies 24, 1–120. Fungi) with special emphasis on the lichen-forming Acarospora- Lutzoni, F., Wagner, P., Reeb, V., Zoller, S., 2000. Integrating ambig- ceae and evolution of polyspory. Mol. Phylogenet. Evol. 32, 1036– uously aligned regions of DNA sequences in phylogenetic analy- 1060. ses without violating positional homology. Syst. Biol. 49, 628– Rehner, S.A., Samuels, G.L., 1995. Molecular systematics of the Hypo- 651. creales: A teleomorph gene phylogeny and the status of their ana- Mason-Gamer, R.J., Kellogg, E.A., 1996. Testing for phylogenetic con- morphs. Can. J. Bot. 73 (Suppl. 1), S816–S823. Xict among molecular data sets in the tribe Triticeae (Gramineae). Rodríguez, F., Oliver, J.F., Marín, A., Medina, J.R., 1990. The general Syst. Biol. 45, 524–545. stochastic model of nucleotide substitution. J. Theor. Biol. 142, May, G.S., Tsang, M.L., Smith, H., Fidel, S., Morris, N.R., 1987. Asper- 485–501. gillus nidulans beta-tubulin genes are unusually divergent. Gene 55 SwoVord, D.L., 2002. PAUP*. Phylogenetic analysis using parsimony (2–3), 231–243. (¤ and other methods). Version 4.0b10. Sinauer, Sunderland, MA. Miller, A.N., 2003. A reinterpretation of the pseudo-bombardioid Tamura, K., Nei, M., 1993. Estimation of the number of nucleotide ascomal wall in taxa in the Lasiosphaeriaceae. Sydowia 55 (2), substitutions in the control region of mitochondrial DNA in 267–273. humans and chimpanzees. Mol. Biol. Evol. 10, 512–526. Miller, A.N., Huhndorf, S.M., 2004a. A natural classiWcation of Lasi- Vilgalys, R., Hester, M., 1990. Rapid identiWcation and mapping of osphaeria based on nuclear LSU rDNA sequences. Mycol. Res. 108 enzymatically ampliWed ribosomal DNA from several Crytococcus (1), 26–34. species. J. Bacteriol. 172, 4238–4246. Miller, A.N., Huhndorf, S.M., 2004b. Using phylogenetic species recog- White, T.J., Bruns, T., Lee, S., Taylor, J., 1990. AmpliWcation and direct nition to delimit species boundaries within Lasiosphaeria. Mycolo- sequencing of fungal ribosomal RNA genes for phylogenetics. In: gia 96, 1106–1127. Innis, M.A. (Ed.), PCR Protocols: A Guide to Methods and Appli- Munk, A., 1953. The system of the Pyrenomycetes. Dansk Bot. Ark. 15 cations. Academic Press, San Diego, pp. 315–322. (2), 1–163. Yoder, A.D., Irwin, J.A., Payseur, B.A., 2001. Failure of the ILD to Orbach, M.J., Porro, E.B., Yanofsky, C., 1986. Cloning and character- determine data combinability for slow loris phylogeny. Syst. Biol. ization of the gene for beta-tubulin from a benomyl-resistant 50, 408–424. mutant of Neurospora crassa and its use as a dominant selectable Zhang, N., Blackwell, M., 2002. Molecular phylogeny of Melanospora marker. Mol. Cell. Biol. 6 (7), 2452–2461. and similar pyrenomycetous fungi. Mycol. Res. 106, 148–155.