Hidden Diversity Behind the Zombie-Ant Fungus Ophiocordyceps unilateralis: Four New Species Described from Carpenter Ants in Minas Gerais, Brazil

Background Ophiocordyceps unilateralis (Clavicipitaceae: Hypocreales) is a fungal pathogen specific to ants of the tribe Camponotini (Formicinae: Formicidae) with a pantropical distribution. This so-called zombie or brain-manipulating fungus alters the behaviour of the ant host, causing it to die in an exposed position, typically clinging onto and biting into the adaxial surface of shrub leaves. We (HCE and DPH) are currently undertaking a worldwide survey to assess the taxonomy and ecology of this highly variable species. Methods We formally describe and name four new species belonging to the O. unilateralis species complex collected from remnant Atlantic rainforest in the south-eastern region (Zona da Mata) of the State of Minas Gerais, Brazil. Fully illustrated descriptions of both the asexual (anamorph) and sexual (teleomorph) stages are provided for each species. The new names are registered in Index Fungorum (registration.indexfungorum.org) and have received IF numbers. This paper is also a test case for the electronic publication of new names in mycology. Conclusions We are only just beginning to understand the taxonomy and ecology of the Ophiocordyceps unilateralis species complex associated with carpenter ants; macroscopically characterised by a single stalk arising from the dorsal neck region of the ant host on which the anamorph occupies the terminal region and the teleomorph occurs as lateral cushions or plates. Each of the four ant species collected - Camponotus rufipes, C. balzani, C. melanoticus and C. novogranadensis - is attacked by a distinct species of Ophiocordyceps readily separated using traditional micromorphology. The new taxa are named according to their ant host.


Introduction
The genus Cordyceps Fr. (1818) was established to accommodate ascomycete fungal pathogens of arthropods bearing the sexual spore (ascospore)-producing structures (ascomata) in or on conspicuous stalks (stromata) arising from the host cadaver. The polyphyletic nature of the genus had been recognised for some time [1], [2] -especially when parasites of truffle-like fungi (Elaphomyces) were included subsequently in the genus -but only recently has this been conclusively confirmed following a multigene study [3]. Species formerly assigned to the genus now occupy four genera in three families of the order Hypocreales.
Ophiocordyceps unilateralis (Tul.) Petch -a ubiquitous pathogen of ants with a pantropical distribution and more commonly known under the name of Cordyceps unilateralis (Tul.) Sacc. -was reinstated in the genus Ophiocordyceps [3], originally erected to accommodate species with non-fragmenting ascospores [1]. This was designated the type genus of the new family Ophiocordycipitacae, and currently comprises around 140 species [3]. It was argued that: ''Because O. unilateralis is a well-known species that was included in the original publication of Ophiocordyceps and because additional Ophiocordyceps species are members of this clade, we apply the name Ophiocordyceps based on the placement of O. unilateralis'' [3]. This species, therefore, is central to our understanding of the taxonomy and functional morphology of this ecologically important group of fungi.
Perversely, the taxonomy of O. unilateralis remains unclear, since apparently, the type specimen is immature and the salient morphological features -ascomata, asci and ascospores -were never included in the original description of Torrubia unilateralis Tul. in 1865 [4]: later confirmed by Y. Kobayasi who examined the holotype held in the Museum of Entomology in Paris [2]. L. Tulasne listed the host as the leaf-cutting ant Atta cephalotes from Brazil, and the specimen is beautifully illustrated by his brother C. Tulasne [4; see Fig. 1a-b]: who is noted for the accuracy of his drawings. Clearly, the ant host depicted is not a leaf-cutter and strongly resembles a carpenter ant, specifically Camponotus sericeiventris Mayr based on the distinctive pronotal plate (Stefan Cover, Museum of Comparative Zoology, Harvard Museum, pers. comm., May 2010; see Fig. 1c). In addition, after more than 150 years of collecting, O. unilateralis has only ever been recorded on ant hosts of the tribe Camponotini, whilst leaf-cutting ants appear to be remarkably free from infections by entomopathogenic fungi with few confirmed associations [5], [6]. Unfortunately, the type specimen is not available for examination since, thus far, it cannot be located at the Paris Museum (Bart Buyck, pers. comm., August 2010). Whether or not the type specimen was actually returned to France from Japan is a moot point since this coincided with the chaotic events of the early stages of World War II. Priority is being given to finding a neotype and on-going searches in Brazil have found two specimens of infected Camponotus sericeiventris, both of which were immature.
We set out to explore the taxonomy, as well as the ecology, of this keystone species because of uncertainties about its status, especially the levels of variation in morphology that had been noted over the years from collections worldwide [1], [2], [4], [7], [8]. Our hypothesis rested on the assumption that this variation was due to geographic or host isolation and that this would probably be best interpreted at the varietal level, as reported previously amongst the 'Cordyceps' species associated with ants [2], [8], [9], [10]. Our 'World Tour' began in Brazil, investigating the pathogens of carpenter ants in the Atlantic rainforest system of the south-eastern region of Minas Gerais State, locally known as the Zona da Mata. Here we report on these initial findings, using 'living' rather than herbarium specimens, that conclusively reject the theory that variation within the so-called O. unilateralis complex is the result of geographic isolation -and, moreover, that variation would involve only minor morphological characters that could be interpreted at the varietal level -and describe four new taxa from different species of Camponotus, all collected within a small area of fragmented, remnant forest. Differences in the functional morphology of all the spore stages within each species reflect the ecology of the ant host, and these are so pronounced that molecular characterisation is not required to separate them. However, DNA has been recovered from the new species and will be sequenced together with planned collections from across the tropics to obtain an overall picture of diversity within O. unilateralis s.l.
Additional Germination process. The majority of ascospores remain unchanged after 28 days on DWA; occasionaly showing cell swellings, especially centrally, with the formation of appressoriallike structures (Fig. 4i).
Anamorph. Hirsutella A-type only, on upper part of stromata: both specimens (see below) bearing only remnant hymenium on ageing stromata.
Additional specimens: Only two specimens collected; one holotype and one isotype deposited.

Summary points
The four new species proposed herein can readily be separated on morphological characteristics (Table 1). In addition, there are marked differences in the germination process with the smaller, more delicate ascospores of O. camponoti-rufipedis and O. camponotinovogranadensis producing secondary spores (capilliconidia); whilst the significantly longer and stouter ascospores of O. camponotibalzani and O. camponoti-melanotici fail to or only occasionaly germinate in vitro, but never produce capilliconidiophores. From these preliminary data, it is tantalising to speculate that each species of the tribe Camponotini may be attacked by a distinct species of Ophiocordyceps. However, no studies, up until now, have examined living material and, therefore, there are few details of patterns of septation (this is difficult to establish in immature or non-discharged spores) and germination, which are considered here as critical additional characters to separate species. Circumstantial evidence can be gleaned from the literature revealing considerable variation in stromatal and ascospore morphology between collections [1], [2], [8], [11], as well as in anamorph morphology [8]. Indeed, the latter study shows a striking range of synanamorphs associated with species of Polyrhachis in Africa. Whether or not this hypothesis is correct will await more in-depth studies using both morphological and molecular data. Some recent collections from North America indicate that this theory may not always hold true but the naming of the new species here, based on the ant host attacked, seems justified and robust enough since there was no evidence of cross infections within overlapping or sympatric ant populations. For this reason, it is considered that the four species described here are distinct from O. unilateralis -even though the holotype is effetebecause it is associated with a different species of Camponotus, purportedly C. sericeiventris. Clearly, however, there is an impasse until mature specimens are found on C. sericeiventris and a neotype is designated.

Functional morphology related to host ecology
We put forward the hypothesis that ant ecology drives the functional morphology of the fungal pathogen. Through a combination of behavioural and morphological studies, we may begin to understand the diversity of O. unilateralis s.l. at the global level which in turn will allow us to look beyond the ant-infecting Ophiocordyceps and ask how specialised are other species of Ophiocordyceps attacking members of diverse insect orders from the Orthoptera to Lepidoptera.

Specimens
Collection of specimens was concentrated in two small areas of remnant, secondary Atlantic rainforest in the south-eastern region of the State of Minas Gerais, Brazil: Mata do Paraíso (MAP), a forest reserve of around 400 ha belonging to the Federal University of Viçosa, ca. 700 m a.s.l.; Parque Estadual de Itacolomi, Ouro Preto (OP), ca. 150 km west of MAP and 1,000 m a.s.l., comprising a much larger area of mixed forest (.7,500 ha), but with collecting restricted to a fragment of low-canopy, seasonally-flooded forest where high densities of infected ants had been recorded previously [12], (Sérvio Ribeiro, pers. comm., May 2010). This involved a careful inspection of shrubs and tree boles up to ca. 2 m and any infected ants plus attached substratum (leaves, small branches and epiphytic lichens) were transferred to sterile containers. Preferably, these were examined the same day for initial identification and the selection of specimens with well developed stromata for morphological studies.

Morphological studies
For those ant species where abundant material was collected, stalks bearing mature stromata were removed from the host, attached at the base to the lid of a plastic Petri dish -previously smeared with a film of Vaseline -and suspended above a dish containing either distilled water agar (DWA) or potato carrot agar (PCA). Where material was in short supply, the infected ant was attached directly to the lid. These were observed daily to monitor release of ascospores, usually signalled by white spore shadows on the agar surface and/or on the lid. Ejected ascospores were transferred aseptically to fresh DWA plates or to PCA, and incubated (20-25uC) for up to one month to observe the germination process, using an Olympus stereomicroscope.
Subsequently, free-hand sections were made from the stromata using a razor blade. For microscopic examination, material was mounted in 1% acid fuchsin and observed using an Olympus BX51 light microscope fitted with a MicroPublisher 3.3 RTV Q an Olympus E330 imaging camera. A minimum of 50 discharged (mature) ascospores were examined and measured for the comparative morphology (Table 1).
Specimens, together with permanent slides of all the relevant structures, were deposited in Herb IMI (RBG, Kew) with isotype collections held in Herb VIC (Federal University of Viçosa, UFV, Minas Gerais, Brazil). Permits for export and collecting were provided by IBAMA to SLE (23920-1). There were no live animals used in our study.

Nomenclature
The electronic version of this document in itself does not represent a published work according to the International Code of Botanical Nomenclature, and hence the new names contained in the electronic version are not effectively published under that Code from the electronic edition alone. Therefore, a separate edition of this document was produced by a method that assures numerous identical printed copies, and those copies were simultaneously distributed (on the publication date noted on the first page of this article) for the purpose of providing a public and permanent scientific record, in accordance with Article 29 of the Code. Copies of the print-only edition of this article were distributed on the publication date to herbaria or generally accessible libraries of the following institutions (CABI, Harvard Farlow, CBS, Kew, New York Botanical Garden, USDA Systematic Mycology and Microbiology Laboratory and Mycology Lab, Beijing). The separate print-only edition is available on request from PLoS (Public Library of Science) by sending a request to PLoS ONE, Public Library of Science, 1160 Battery Street, Suite 100, San Francisco, CA 94111, USA along with a check for $10 (to cover printing and postage) payable to ''Public Library of Science''. In addition, new names contained in this work have been submitted to Index Fungorum (http://www.indexfungorum.org/), from where they will be made available to the Global Names Index.