Mycol. Res. 108 (1): 26–34 (January 2004). f The British Mycological Society 26 DOI: 10.1017/S0953756203008864 Printed in the United Kingdom. A natural classification of Lasiosphaeria based on nuclear LSU rDNA sequences Andrew N. MILLER1,2 and Sabine M. HUHNDORF2 1 University of Illinois at Chicago, Department of Biological Sciences, Chicago, Illinois 60607-7060, USA. 2 Botany Department, The Field Museum of Natural History, Chicago, Illinois 60605-2496, USA. E-mail : amiller@fieldmuseum.org Received 9 July 2003; accepted 6 October 2003. The current circumscription of Lasiosphaeria includes taxa with a wide variety of ascomatal walls, ascomatal wall vestitures, and ascospore morphologies and a broad range of putative anamorphs. Despite the complexity of morphological characters in the genus, species within Lasiosphaeria can be arranged into four groups based on ascospore morphology. Taxa which possessed ascospores in each of the four groups were used in phylogenetic analyses of partial nuclear large subunit (LSU) rDNA sequences to test the monophyly of the genus and determine relationships among its species. Lasiosphaeria was found to be highly polyphyletic in that species segregated into seven well-supported monophyletic clades dispersed among several orders. Three new genera, Echinosphaeria, Hilberina, and Immersiella, are erected for three of these clades while the genus Lasiosphaeris is reintroduced for a fourth clade. These data support Ruzenia as a previously established genus and the transfer of Lasiosphaeria raciborskii to Chaetosphaeria. The circumscription of Lasiosphaeria has been considerably narrowed to better reflect a natural classification. These taxonomic changes are additionally supported by a combination of morphological characters which are discussed in relation to the phylogenetic trees. to obpyriform, glabrous, tomentose, villose, or setose, INTRODUCTION and possess an outer ascomal wall that is smooth or The genus Lasiosphaeria is one of the most speciose in tuberculate. Ascospores can be hyaline to yellowish and the Sordariales. It is composed of primarily lignicolous usually turn pale brown with age, are allantoid, reni- taxa which are commonly found throughout temperate form, cylindrical, filiform, or fusiform, may bend or regions. The genus was originally based on seven species taper to a point in the lower third, are aseptate to multi- along with five additional species, which were presum- septate, and may possess hyaline, gelatinous append- ably considered to belong in the genus, but were ages or mucilaginous tips at one or both ends. Species not transferred (Cesati & De Notaris 1863). Four of of Lasiosphaeria are also associated with a broad range these species (L. ovina, L. hirsuta, L. rhacodium, and of anamorphs ranging from Phialophora in L. hirsuta L. canescens) currently remain in the genus. The history (Hughes 1951, Gams & Holubova´ -Jechova´ 1976), L. of Lasiosphaeria and its synonyms is quite complex and hispida (Gams 2000), and L. ovina (Gams & Holubova´ - has already been elaborately discussed by Lundqvist Jechova´ 1976), to Endophragmiella and Selenosporella- (1972), who also explained why L. ovina should be rec- like synanamorphs associated with L. canescens and ognized as its type species. Since its conception, the L. punctata (Hughes 1979, Sivanesan 1983), to an un- genus has continually been expanded to include an identified anamorph with blastic-annellidic conidio- array of taxa with a broad range in ascomatal walls, genesis associated with L. dactylina (Hilber, Webster & ascomatal wall vestitures, and ascospore morphologies. Hilber 1987). Lasiosphaeria currently contains over 180 names (Reed The generic concept of Lasiosphaeria has tradition- & Farr 1993; Index of Fungi), most of which probably ally been overly broad and several workers have dis- do not belong, and a monographic study to revise cussed the complexity of the genus (Seaver 1912, Munk all type material is presently underway (Miller & 1957, Carroll & Munk 1964, Lundqvist 1972, Hilber Huhndorf, unpubl.). The generic concept in its broadest & Hilber 1979, Candoussau, Fournier & Magni 2001, sense contains taxa with large, membraneous ascomata Hilber & Hilber 2002). Although definite criteria that are immersed to erumpent, or superficial, globose for placing taxa into Lasiosphaeria have never been A. N. Miller and S. M. Huhndorf 27 Figs 1–9. Representative species of Lasiosphaeria s. lat. divided into four groups (A–D) based on ascospore morphology: Figs 1–2=Group A; Figs 3–4=Group B; Figs 5–6=Group C; Figs 7–9=Group D. Bars=10 mm. Fig. 1. Ruzenia spermoides (SMH4606). Fig. 2. Echinosphaeria canescens (SMH2627). Figs 3–4. Chaetosphaeria raciborskii (SMH2499, SMH2132). Figs 5–6. Hilberina caudata (SMH3156, SMH3918). Fig. 7. Immersiella immersa (SMH4104). Fig. 8. Lasiosphaeris hispida (SMH3336). Fig. 9. Lasiosphaeria ovina (SMH1538). established (Carroll & Munk 1964, Candoussau et al. characters are phylogenetically informative for de- 2001), taxa have generally been added or transferred limiting Lasiosphaeria. into the genus based on overall similarities in their ascomatal walls, ascomatal wall vestitures, and/or ascospore morphologies. Carroll & Munk (1964) MATERIALS AND METHODS pointed out that ‘radically different spore types may Taxon sampling occur together with one and the same peridium type’. However, despite the wide range of ascospore types Taxa sequenced in this study are listed in Table 1 along currently accepted in Lasiosphaeria, taxa can be ar- with their source information, geographical locality, ranged into four generalized groups (Figs 1–9, Groups and GenBank accession number. Representatives from A–D) based on their ascospore morphologies as several families and orders within the Sordariomycetes was previously shown by Candoussau et al. (2001). were included to determine the phylogenetic position Although several species have been transferred to other of putatively polyphyletic taxa currently included in genera over the years (cfr Candoussau et al. 2001), a Lasiosphaeria. Four members of the Xylariales, Daldi- number of new species with a variety of ascomal and nia concentrica, Diatrype disciformis, Eutypa sp., and ascospore morphologies have recently been added Xylaria hypoxylon, were used as outgroups. All voucher (Re´ blova´ 1997, Vasilyeva 1998, Candoussau et al. 2001, specimens are deposited in the Field Museum Mycology Taylor et al. 2001). The circumscription of Lasio- Herbarium (F). Additional taxa obtained from Gen- sphaeria will continue to expand and become increas- Bank are listed in Table 2. Cultures of multispore ingly heterogeneous as additional taxa are added until a isolates were obtained following the techniques of clear delimitation is provided. Therefore, the objective Huhndorf, Miller & Ferna´ ndez (2004). Ascomata were of this study was to use an independent data set in squash-mounted in water and images of micromorpho- the form of partial nuclear LSU rDNA sequences to: logical structures were captured with a Dage DC-330 (1) test the monophyly of Lasiosphaeria; (2) assess video system mounted on a Zeiss Axioskop microscope relationships among species of Lasiosphaeria posses- using differential interference (DIC) microscopy. Im- sing different ascomal, ascospore, and anamorph ages were processed using Adobe Photoshop 3.0 or 5.0 morphologies; and (3) determine which morphological (Adobe Systems, Mountain View, CA). A natural classification of Lasiosphaeria 28 Table 1. Taxa sequenced in this study. subjected to a symmetric stepmatrix generated using STMatrix ver. 2.2 (Franc¸ ois Lutzoni & Stefan Zoller, GenBank Geographical accession Biology Department, Duke University). This program Taxon Source1 locality nos. calculates the costs for changes among these character states based on the negative natural logarithm of the Chaetosphaeria SMH3119 Puerto Rico AY436402 percentages of reciprocal changes between any two raciborskii Echinosphaeria SMH4791 (a)2 Indiana AY436403 character states. The phylogenetic signal from twelve of canescens the thirteen ambiguous regions was recovered using E. canescens TL5730 (a) Denmark AY436404 INAASE (Lutzoni et al. 2000) and also analyzed in the Hilberina caudata SMH2107 Puerto Rico AY4364053 unequally weighted analyses. The remaining ambiguous H. caudata SMH3918 Michigan AY436406 region was excluded because its recoded character Immersiella caudata SMH3298 North AY4364073 Carolina contained more than 32 character states, which is not I. immersa SMH2589 Michigan AY4364083 allowed in PAUP* 4.0b10. Branch support for all MP I. immersa SMH4104 Wisconsin AY436409 analyses was estimated by performing 1000 bootstrap Lasiosphaeria glabrata TL4529 (a) Denmark AY436410 replicates (Felsenstein 1985) with a heuristic search L. glabrata SMH4617 Denmark AY436411 consisting of 100 random addition replicates for each L. lanuginosa SMH3819 North AY436412 Carolina bootstrap replicate using the above settings. L. ovina SMH4605 Denmark AY436413 The best-fit model of evolution was determined by L. rugulosa SMH1518 Puerto Rico AY4364143 MODELTEST 3.06 (Posada & Crandall 1998). While L. sorbina GJSL555 Louisiana AY4364153 maximum likelihood analyses were not compu- L. sorbina CBS885.85 Germany AY436416 tationally possible due to the large size of the data set, Lasiosphaeris hirsuta SMH1543 Wisconsin AY4364173 L. hirsuta JF02183 (a) France AY436418 analyses of Bayesian inference were performed using L. hispida SMH3336 North AY4364193 MrBayes 3.0b4 (Huelsenbeck & Ronquist 2001) (http:// Carolina