Insights into the Ecology and Evolution of Blattodea-associated Ophiocordyceps

Entomopathogenic fungi are ubiquitous inhabitants of forests worldwide, remarkably in tropical regions. Among these fungi, one of the most abundant and diverse is the genus Ophiocordyceps. These fungi are particularly diverse and more commonly found parasitizing coleopteran, lepidopteran, hymenopteran and hemipteran insects. However, other insect orders are also parasitized by these fungi, for example the blattodeans (termites and cockroaches). Despite their ubiquity in nearly all environments insects occur, blattodeans are rarely found infected by lamentous fungi and thus, their ecology and evolutionary history remains obscure. In this study, we propose a new species of Ophiocordyceps infecting the social cockroaches Salganea esakii and S. taiwanensis, based on 16 years of collections and eld observations, especially the Ryukyu Archipelago. We found a high degree of genetic similarity between specimens from different islands, infecting two Salganea species and that this relationship is ancient, likely not originated from a recent host jump. Furthermore, we found that Ophiocordyceps lineages infecting cockroaches evolved around the same time, at least twice, one from beetles and the other from termites. We have also investigated the evolutionary relationships between Ophiocordyceps and termites and present the phylogenetic placement of O. blattae for the rst time. present, rst B. teleomorph roach on S. esakii collected in Yakushima on June 17, 2017 and a closeup illustration of perithecia. C-E. distinct developmental stages of O. salganeicola on S. taiwanensis, collected in Katsuu-dake, Nago-shi, Okinawa Apr 23, 2016 D. Mori04 and E. Kunigami-son, Okinawa on May 29, 2016.


Introduction
The genus Ophiocordyceps is composed almost entirely of insect parasites, with few examples of bene cial endosymbionts of sap-sucking hemipterans (Quandt et al. 2014, Gomez-Polo, et al. 2017, Matsuura et al. 2018). The genus was erected by Petch (1931) to accommodate species of Cordyceps exhibiting clavate asci containing spores that do not disarticulate into partspores, contrasting with "the majority of the species of Cordyceps which have been described" at that time, exhibiting cylindrical asci and spores that readily disarticulate into numerous short partspores upon maturity. The diversity of Ophiocordyceps has been increasingly unraveled in the last decade, especially with discoveries of species associated with Hymenoptera, Lepidoptera and Hemiptera (Araújo et  parasitism of microsporidians and protozoans on these insects is relatively common and known for more than a century (Crawley 1905, Woolever 1966, Purrini et al. 1988 while O. blattarioides exhibit a stalk bearing a globoid fertile part at the tip, producing spores that disarticulate into partspores of 6-12 × 1.5 µm. The host death location is also distinct with O. blattae killing its hosts on the underside of leaves while O. blattarioides is found buried in the leaf litter. Generally speaking, there is not su cient information on the evolution and ecology of cockroach-associated entomoparasitic fungi. In this study, we aim to propose a new species of Ophiocordyceps that parasitizes two social woodfeeding cockroach species distributed in southwestern part and Nansei islands of Japan, both living inside decaying logs. We provide morphological, molecular and ecological data to support the new species proposal with insights into the evolutionary origins of the parasitic fungi of cockroaches and their close related termites. We also present, for the rst time, the phylogenetic position of O. blattae. blattae from Thailand used in this study (the actual specimen), with two ascomata arising laterally on both sides.

Materials And Methods
Sampling Surveys were undertaken in the Japanese warm temperate and subtropical evergreen forests mainly consisting of trees belonging to Fagaceae, Lauraceae and Theaceae in Kunigami-son, Okinawa, Yakushima, Kagoshima and Nobeoka, Miyazaki. The parasitized cockroarch samples of two host species, namely Salganea esakii and taiwanensis, were mainly collected in the small humid valley or riparian forests where Castanopsissieboldii, Distylium racemosum and Sche era heptaphylla grow, but also in the secondary forest harboring Alnus japonica after the deforestation of Castanopsis in Okinawa. The specimens used in this study were always found hidden inside soft rotten logs or large fallen branches of C.sieboldii. A. japonica, and so forth, with only the fungus sprouting out (Fig. 2). The infected cockroaches, and the substrata they were attached to, were collected in plastic containers and transported to the laboratory. Some specimens were investigated immediately after the collection, while others were dried for many years before analyzed. The specimens were photographed individually, using a Canon 7D camera, equipped with an EF-100 mm macro lens or a MP-E 65 mm (5X) lens with a MT-24EX Canon macro lite ash attached.

Morphological studies
For macro-morphological characterization, specimens were examined using a stereoscopic microscope (Leica S8 APO) and sorted for further micro-morphological investigation. The characters investigated were ascomatal size, colour, position, presence/absence and characterization of asexual morphs and perithecial insertion (e.g. immersed, semi-immersed, erumpent, super cial). For micro-morphological characterization, either free-hand or cryo-sectioning of the ascoma was performed using a Leica CM1850 Cryostat. Samples were mounted on a slide with plain lactic acid or lacto-fuchsin (0.1g of acid fuchsin in 100 mL of lactic acid) for light microscopy examination using an (Nikon Eclipse Ni-U). A minimum of 50 ascospores were measured for morphological comparison. The illustrations of fungal specimens were drawn based on the observation of photographs using drawing pens 0.13 mm and 0.2 mm (Rotring), painted by watercolors (HOLBEIN Art Materials Inc.) and scanned for imaging (Fig. 3). We also present the morphological comparison between Ophiocordyceps species infecting cockroaches and termites (Table 1).

DNA extraction, PCR and sequencing
All specimens used in this study were collected in their natural habitat. The material was preserved either dried or in ethanol and DNA extractions were performed with the following protocol: Parts of fungal tissues were removed from the host, placed in 1.5 ml Eppendorf tubes with 100-200 µl of CTAB readily after its collection and stored at a room temperature, or entire samples were immersed in 70% ethanol and stored in the freezer. For DNA extraction, the samples were ground mechanically, added 400 µl of CTAB and incubated at 60ºC for 20 min and centrifuged for 10 min at 14,000 rpm. The supernatant (approx. 400 µl) was transferred to a new 1.5 ml Eppendorf tube, mixed with 500 µl of 24:1 Chloroform: Isoamyl-alcohol (Sigma) and mixed by inverting. The mix was then centrifuged for 20 min at 14,000 rpm and the supernatant transferred to a new 1.5 ml Eppendorf tube and further cleaned using the GeneCleanIII kit (MP Biomedicals), following the recommended protocol. The only step modi ed was the addition 30 µl of GlassMilk per sample, instead of the recommended 10 µl, aiming to increase yield.
Five loci were used in the analyses, i.e. small subunit nuclear ribosomal DNA (SSU), large subunit nuclear ribosomal DNA (LSU), translation elongation factor 1-α (TEF) and the largest and second largest subunits of RNA polymerase II (RPB1 and RPB2 respectively) with a total read length of 4,815 bp. The primers used were, SSU: 82F (5'-GAAACTGCGAATGGCT-3') and 1067R (5' min in the thermocycler. The processed PCR products were directly sequenced by a capillary DNA sequencer, Genetic Analyzer 3130xl (Applied Biosystems) at C-RAC of the University of the Ryukyus.

Phylogenetic analyses
The raw sequence reads (.ab1 les) were edited manually using Geneious 11.1.5 (https://www.geneious.com). Individual gene alignments were generated by MAFFT (Katoh & Stanley 2013). The alignment of every gene was improved manually, annotated and concatenated into a single combined dataset using Geneious 11.1.5 (https://www.geneious.com). Ambiguously aligned regions were excluded from phylogenetic analysis and gaps were treated as missing data. The nal alignment length was 4,629 bp: 1,020 bp for SSU, 870 bp for LSU, 967 bp for TEF, 683 bp for RPB1 and 1,089 for RPB2. Maximum likelihood (ML) analysis was performed with RAxML version 8.2.4 (Stamatakis 2006) on a concatenated dataset containing all ve genes. The dataset consisted of 11 data partitions, 2 for SSU and LSU, and 9 for each codon position of the three protein coding genes: TEF, RPB1 and RPB2. The GTRGAMMA model of nucleotide substitution was employed during the generation of 1,000 bootstrap replicates. The sequences for all Ophiocordyceps used in this study are presented in Table 2.

Collections
We collected 26 samples in two locations in Okinawa, 27 in Yakushima isl., Kagoshima, and 5 in Miyazaki. Four specimens from Yonahadake and Kunigami (Okinawa) and one from Kagoshima were used for DNA extraction and sequencing for each prefecture (see Fig. 4, Table 3). Almost all fungal specimens were collected from adults of S. esakii and S. taiwanensis between April -June from 2004 to 2019.    6. Ophiocordyceps salganeicola on Salganea esakii (fresh specimen) from Yakushima, Kagoshima (JPMA106). a.Salganea esakii with a single robust ascoma; b. Close-up showing early stage ascoma arising from ventral pronotum; c-d. Close-up of ascoma; e. Cross-section of ascoma; f-g. Perithecia; h. Ascospores within Ascus; i-j. 8-celled ascospores.

Molecular phylogeny and Evolutionary origins of cockroach-associated Ophiocordyceps
We obtained 20 new sequences from 5 specimens of O. salganeicola (Fig. 4, Table 3). Our phylogenetic analysis is in accordance with previously published Ophiocordyceps topology (Quandt et  All O.salganeicola specimens we collected, from different parts of Japan and infecting two species of Salganea, clustered together as a single species with considerably high degree of genetic similarity with a long branch (Fig. 7). It formed a monophyletic group with another cockroach-associated species, O. blattae, which is the type species for Ophiocordyceps. This is the rst time O. blattae is included in a phylogenetic study.
Our results indicate that Ophiocordyceps originated from a beetle-associated ancestral (72% ACSR), corroborating with previous studies (Araújo & Hughes 2019). For the cockroach parasites, we found at least two independent origins within Ophiocordyceps, one within Paraisaria clade, i.e. Paraisaria blattarioides (Fig. 7 node A), and the other within the hirsutelloid species, i.e. O. salganeicola sp nov and O. blattae (Fig. 7 node B). The ancestral host association for the cockroach-associated Paraisaria lineage was ambiguously recovered, while for the hirsutelloid cockroach-associated species our data show it has originated likely from a termite-associated ancestor, although not strongly supported (44% ACSR). We also found that the association with termites is older than cockroaches, evolving independently at least twice ( Fig. 7 nodes C and D). The oldest, would have arisen from beetles to termites (65% ACSR, Fig. 7 node C). However, the origins of O. brunneirubra remains uncertain as part of the ancient termiteassociated lineage (Fig. 7 node C) or if it jumped more recently from Hymenoptera to termites (Fig. 7 Node C).
Paraisaria clade is an ecologically heterogeneous group composed by species parasitic on Coleoptera, Orthoptera, Lepidoptera and Hemiptera (Mongkolsamrit et al. 2019). Our ACSR analysis provided weak resolution for the origins of O. amazonica/O. blattarioides/O. gracilis clade with 50.1% for Orthoptera, 25.9% for Blattodea (cockroaches) and 10.9% for Lepidoptera (Fig. 7 Node A). Our data also did not provide strong support for the ancestral of O. blattarioides with 51.1% for Blattodea (cockroaches), 21% for Orthoptera and 20.6% for Lepidoptera (Fig. 7 Node A). Nevertheless, the whole Paraisaria lineage was strongly supported as evolved from a beetle parasite (Fig. 7 Node F, ACSR = 81.1%).
On the other hand, for the novel clade composed by O. salganeicola sp. nov. and O. blattae, our results suggest (BS=87; ACSR=72.4%) that it evolved from an ancestral parasite on termites (Fig. 7 -Node E). Ophiocordyceps salganeicola/blattae was retrieved as a sister group to a clade composed mostly by termite (Blattodea, Termitidae) parasites with also species associated with hemipterans (Pseudococcidae) and mites (Acari, Eriophyidae). According to our results, all host switches in this clade occurred from termites (i.e. termites to Coleoptera, termites to Hemiptera, termites to Acari and termites to cockroaches). Unexpectedly, our analyses also suggest (ACSR=61.7%) the clade composed by parasites of Coleoptera, Lepidoptera and Hemiptera, including the economically and culturally important O. sinensis, could have been originated from an ancestor infecting termite, instead of beetle larvae as previously proposed (Araújo & Hughes 2019). . This is particularly true for colonial cockroaches that spend most of their lives protected inside nests, for example some wood-feeding species within the families Cryptocercidae and Blaberidae, speci cally the subfamily Panesthiinae (Panesthiini, Ancaudeliini, Caepariini and Salganeini). Among those groups, one of the most well-known social cockroaches is the genus Salganea, comprised by about 50 spp. (Beccaloni 2007, Bell et al. 2007, Wang et al. 2014). All the known species within the genus live within and feed on decaying wood, building chambers and galleries inside hardwood or coniferous logs that may take decades to degrade (Maekawa et al. 2008), providing long-term stable homeostatic conditions. Such protected environment certainly bene ts indirectly the fungal parasites that are already inside the host body. An exposed cadaver on the forest oor would be much more susceptible to be eaten by scavengers or being consumed by other microorganisms, which would naturally antagonize with the fungal growth.
Salganea species form social groups, composed mostly by biparental families, consisting of a malefemale and their offspring (Maekawa et al. 2008). Sociality endows insects with advantages such as increased e ciency of brood care, foraging and anti-predator defenses. However, infectious diseases can potentially spread more easily within a colony because of their high densities, frequent social contact and also that group members are often close relatives and thus susceptible to the same parasitic infections (Cremer et al. 2007). Therefore, it is surprising that only three species of Ophiocordyceps, a common and widespread group of entomopathogenic fungi, were recorded infecting the equally diverse and globally distributed cockroaches (Bourguignon et al. 2018). Ophiocordyceps species infecting social insects, notably ants, are one of the most broadly distributed and ubiquitous entomopathogenic fungi in tropical forests worldwide (Araújo et al. 2015(Araújo et al. , 2018. They often form epizootic events, in which hundreds of infected ants can be found a small patch of forest (Evans and Samson 1982, Pontoppidan et al. 2009).
On the other hand, however, Ophiocordyceps on social cockroaches are rare in Japan and only one or two infected individuals are collected in the same log, despite the ubiquity and abundance of hosts in one area.
Based on our extensive eld surveys we found that fruiting bodies of O. salganeicola in Okinawa start to emerge in decaying logs in early April. However, they seem to require at least a few months to be mature and develop the sexual morph in the eld. This development occurs in parallel with the mating season of the host cockroaches from April to July when newly emerged adults leave their logs and parents, y, mate and burrow into a new nest (Osaki Haruka, personal communication). Presumably, these young adults might become infected by the ascospores/conidia of O. salganeicola during colonization of a new log. The host is then later killed and consumed by the fungal parasite, eventually producing new fruiting bodies in the next mating season. There has been no record of an infected nymph out of more than 26 fungal specimens observed in Okinawa, but only a single infected nymph (see Fig. 2

Behavior manipulation
The behavior manipulation caused by Ophiocordyceps fungi on their hosts is a striking phenomenon, especially on species associated with ants, the so-called "zombie-ant fungi" (Andersen et al. 2009, Araújo et al. 2018). It has been posited that species within the Ophiocordyceps unilateralis core clade infecting Camponotini ants evolved such an ability as a response to the strong social immunity displayed by ant societies that prevents fungal transmission and development inside the colony (Araújo & Hughes 2019). Conversely, as far as we know, there is no evidence of social cockroaches recognizing the infected members of their colony, except for the parental and sibling's grooming behavior that might fend off super cial parasites. Thus, fungal infection, development and transmission can potentially occur in the same log where other members of the colony still inhabit, implying no drastic behavior manipulation is needed in order to remove the host from its nest and thus complete the parasite's life-cycle. However, there is a possibility of a subtle manipulation.
Salganea cockroaches inhabit deep inside the trunk, just becoming exposed to the external environment in the mating season, whereas the Ophiocordyceps-infected ones are found only few centimeters below the wood surface (Fig. 2). Thus, we could speculate that the fungus might potentially be able to manipulate host's behavior by leading to a migration towards a more super cial layer inside the log. This host migration could be stimulated by the need of a more oxygen-and water-rich stratum and/or attraction to the light coming from an opening, through which the fungus sprout outwards to further shoot its spores (Fig. 2, A-B). Future studies could test this hypothesis and help us to clarify such a speculation.

Host Association, Speciation and Distribution
While some distinct -mostly macro -morphological features can be observed on O. salganeicola infecting both host species that diverged from a single ancestor only around 4-5 mya (Maekawa et al. 1999, Maekawa & Matsumoto 2003, nucleotide sequences are highly conserved among diverse strains in wide range of geographic regions and islands. The number of polymorphic sites within all ve gene sequences from multiple samples was only 2 out of 4,202 and both were synonymous. Thus, geographic and reproductive isolation of the fungal strains may have not yet resulted in the allopatric speciation of O. salganeicola. The long branch length and high host speci city to the genus Salganea indicate a long coevolutionary relationship with the host populations and unlikelihood of recent host jumping (Fig. 7, Node  B). However, we still do not know whether the single parasite strain can only persist in one geographic region and/or island infecting the same host populations over generations, or jump across multiple closerelated host populations and species horizontally even after such a long geographic isolation of the Ryukyu islands. In some of these islands such as Amamioshima and Tokunoshima in Kagoshima and Ishigaki-jima and Iriomote-jima in Okinawa, there has been no collection record of O. salganeicola, suggesting the loss of entomoparasite strains in these host populations. These questions on the evolution of host speci city and associations of O. salganeicola in the Japanese archipelago of Ryukyus deserve particular attention for studying host-parasite co-evolution in the context of island biology, which can be possibly be tested by the eld-collected fungal ascoma and laboratory-reared Salganea colonies from different islands in future studies.

Conclusion
The Japanese fungal community, remarkably the entomopathogens, likely harbors one of the largest reservoirs of undocumented fungal species in the world. We still know very little about these organisms and their ecological roles in the environment and dynamic associations with host insects, including blattodean-associated ones such as O. salganeicola. Therefore, biodiversity studies are crucial to understand their true diversity with the invaluable help by amateur and professional mycologists. As we move forward and describe more and more species through microscopic and molecular tools, new insights into the evolutionary origins of these organisms are revealed, as well as their ecological associations with the insect hosts. Currently, there is still a substantial gap on the knowledge about insect ecology and fungal biology along the context of host-parasite interactions and their life cycles. In this study, our goal was to describe an ecologically rare fungal species parasitizing unique social insects, and to provide some insights into their evolution by considering natural histories of both the parasite and its host. Thereby, we expect that this study will contribute to the understanding of one of the most proli c and diverse groups of entomopathogenic fungi, the genus Ophiocordyceps. Figure 1 Ophiocordyceps blattae illustrations. a. the holotype original illustration of the host with fungus arising laterally and ascus with septate ascospores (adapted from Petch 1924); b. Illustration of O. blattae from Thailand used in this study (the actual specimen), with two ascomata arising laterally on both sides.