Microbes Environ. Vol. 19, No. 3, 215–220, 2004 http://wwwsoc.nii.ac.jp/jsme2/

Comparison of Symbiotic Flagellate Faunae between and a Wood-Feeding of the

OSAMU KITADE1*

1 Natural History Laboratory, Faculty of Science, Ibaraki University, Mito, Ibaraki 310–8512, Japan

(Received May 22, 2004—Acccepetd July 1, 2004)

Termites of most isopteran families and wood-feeding of the genus Cryptocercus usually harbor more than one symbiotic flagellate in their hindgut. To evaluate the similarity of their symbiont faunae, data on symbiont composition at a generic level were examined by cluster analysis and type III quantification method. In both analyses, the symbiont composition recorded from host belonging to the same families or monophyletic family groups tended to be similar. This tendency was particularly remarkable in the clade Kalo- and the clade plus . Two basal host groups, the Cryptocercidae and the Mastotermitidae, exhibited very different symbiont compositions. These findings suggested that the symbiont faunae mainly reflect the host’s phylogenetic relationships. Within the Rhinotermitid hosts, the genus Reticuliter- mes showed a unique symbiont fauna although it is not a basal taxon in the Rhinotermitidae. Horizontal transfers of symbiotic protists might explain such anomalistic fauna.

Key words: cluster analysis, community, cospeciation, Oxymonadea, Parabasalea, protist, type III quantification method

Termites (Isoptera) are the most ecologically important era Rhinotermes, Parrhinotermes, Termitogeton16) to 26 in wood-feeding because of their huge biomass in the the wood-feeding cockroach Cryptocercus28). The composi- tropics and their large contribution to carbon cycling in ter- tion of symbiont species is usually a host species, which restrial ecosystems. Lower termites, representing six of the may be relevant to the mode of symbiont transmission. That seven families in the Isoptera, have a symbiotic protist com- is, the symbionts are transmitted between host individuals in munity in their hindgut9,10). These symbionts belong to basal a colony by way of proctodeal trophallaxis, and consequent- eukaryotic taxa, flagellate orders of the phylum Axostylata: ly a newly founded host colony should succeed the Trichomonadida, Hypermastigida (class Parabasalea) and symbiont faunae of the mother nests of its king and queen. Oxymonadida (class Oxymonadea)7). Recently, Lo et al.18) Because the host-symbiont relationship is obligatorily suggested that the cockroach order is not mono- mutualistic, the protist faunae should reflect the phylogenet- phyletic, but a cockroach genus, Cryptocercus (Cryptocer- ic relationships of the hosts. Some authors have discussed cidae), is a sister group of the order Isoptera. Spe- infection patterns of particular symbiont taxa in connection cies of Cryptocercus are all wood-feeding and possess with phylogenetic relationships of host termites9,14,15). protist symbionts of the same orders as those of termites4). However, these arguments were not convincing due to the Cellulases produced by the symbionts are essential for limited amount of available data. In this study, I present the the host termites to digest wood-fibers10) and the symbionts results of a composition of symbiotic protist fauna between are dependent on their hosts for food supply and anaerobic most termite genera and the wood feeding cockroach genus habitats. The number of symbiont species in a host’s hind- Cryptocercus. gut varies among host taxa, such as from one in termite gen- Materials and Methods * Corresponding author; E-mail: [email protected], Tel: 81–29–228–8375, Fax: 81–29–228–8404 The data on the symbiont composition were taken mainly 216 KITADE

Table 1. Number of symbiotic flagellate genera found in each host genus

Host taxa No. symbiont genera No. host species examined* T** H** O** Total Order BLATTODEA Family Cryptocercidae Genus 1. Cryptocercus 4 (9) 0 11 6 17 Order ISOPTERA Family Mastotermitidae Genus 2. 1 (1) 2 2 0 4 Family Genus 3. 15 (45) 14 0 3 17 4. Epicalotermes 5 (6) 2 1 0 3 5. Bifiditermes 7 (13) 4 1 1 6 6. Proneotermes 1 (2) 2 1 1 4 7. Allotermes 3 (3) 3 1 0 4 8. 1 (2) 2 1 1 4 9. 15 (27) 12 2 1 15 10. 1 (1) 1 1 1 3 11. Postelectrotermes 11 (15) 5 4 2 11 12. 26 (109) 14 4 1 19 13. Rugitermes 5 (13) 5 1 1 7 14. 10 (18) 6 2 3 11 15. Paraneotermes 1 (1) 5 4 1 10 16. Gryptotermes 29 (120) 9 1 3 13 17. Calcaritermes 6 (13) 6 0 2 8 Family Genus 18. Anacanthotermes 7 (10) 6 7 1 14 19. Microhodotermes 2 (3) 2 2 1 4 20. Hodotermes 1 (2) 2 3 0 5 Family Genus 21. Archotermopsis 1 (1) 3 4 1 8 22. Hodotermopsis 1 (1) 2 8 2 12 23. Zootermopsis 3 (3) 3 1 1 5 24. Porotermes 3 (3) 3 4 1 8 25. Stolotermes 7 (7) 2 5 0 7 Family Rhinotermitidae Genus 26. 2 (7) 0 4 0 4 27. 22 (75) 4 6 2 12 28. 10 (38) 3 5 0 8 29. 14 (71) 3 3 0 6 30. 3 (18) 0 4 0 4 31. Termitogeton 1 (3) 0 1 0 1 32. 5 (32) 3 5 0 8 33. Rhinotermes 3 (4) 0 1 0 1 34. Dolichorhinotermes 1 (5) 1 1 0 2 35. Parrhinotermes 4 (9) 0 1 0 1 Family Serritermitidae Genus 36. Glossotermes 1 (1) 1 2 0 3 37. Serritermes 1 (1) 1 2 0 3 *: Number of known host species (data from Kambhampati and Eggleton13)) in each genus is also shown in brackets. **: T, Order Trichomo- nadida; H, Order Hypermastigida; O, Order Oxymonadida. Flagellate Faunae of Termites 217 from Yamin28), but also from taxonomic/ecological 1,16,20,21,22) papers . Symbiont data for the following host spe- Results cies were obtained by direct observation: Hodotermopsis Cluster Analysis based on the similarities of symbiont sjoestedti, Porotermes planiceps (Termopsidae); Micro- composition hodotermes viator (Hodotermitidae); Glyptotermes satsum- ensis, G. fuscus, G. nakajimai (Kalotermitidae); Serriter- The analysis using r as similarity (Figure 1) divided host mes serrifer, Glossotermes oculatus (Serritermitidae); genera into six clusters (I–VI) at a distance level of 0.1, and Rhinotermes marginalis, R. hispidus, Dolichorhinotermes the genera in clusters IV and V each into subclusters (IVa, sp. (Rhinotermitidae); and Cryptocercus kyebangensis IVb, Va, Vb) at a similarity level of 0.2. The result general- (Cryptocercidae). I followed Kambhampati and Eggleton13) ly coincided with the current classification of the host gen- for the classification of host insects except for the familial era at the family level. Fourteen out of the 15 genera in the assignment of the Glossotermes to the Serritermitidae2). The Kalotermitidae and four of the five genera in the Termop- family Termitidae lack symbiotic protists, and were not in- sidae formed exclusive clusters (III and V, respectively). cluded in the analyses. As the species level of Nine out of the ten genera in the Rhinotermitidae also both hosts and symbiotic protists is still poorly conducted, formed an exclusive cluster, VI, together with two genera in the data were combined at the generic level for both hosts and symbionts. Host genera for which less than three sym- biont species have been recorded were not included in the data set, except for four Rhinotermitid genera for which I directly examined more than one colony. For the protist tax- onomy I followed Yamin28), but treated Spirotrichonympha, Spironympha, Pyrsonympha and Dinenympha as indepen- dent genera as in Grassé6) (Table 1). Cluster analyses of the host genera using the unweighted pair group method using arithmetic averages (UPGMA)23) were made based on similarities either in the symbiont com- position or in geographic distribution. For the symbiont composition, I calculated two similarity indices based on the presence/absence (1/0) data of each symbiont genus: the correlation coefficient (r) and Jaccard’s coefficient11). To quantify the similarity in biogeographical affiliation, I calculated the correlation coefficient between the host gen- era based on a matrix representing the presence/absence (1/ 0) data of each host genus in each biogeographical region. For the system of biogeographical regions, I followed Kam- bhampati and Eggleton13) in which the world was divided into the following nine regions: Nearctic, West Palearctic, East Palearctic, Neotropical, Afrotropical, Malagasy, Oriental, Papuan, and Australian regions. Calculation of the similarity indices and clustering was conducted with the STATISTICA (StatSoft Inc., Tulsa, OK) and the R 4.0d63) program packages. Type III quantification method8) was also used to analyze the similarity patterns of the symbiont composition using the “Let’sStat! Pro” package (provided by M. Kitamura). Fig. 1. Results of UPGMA clustering of host genera based on the The presence/absence data of symbiont genera were treated similarity of symbiont composition using the correlation coeffi- as “category (1/0) data” and directly subjected to an analy- cient (r) as a similarity index. Symbols and numerals (see Table 1) indicate host families and genera, respectively. Capital letters sis without assigning dummy variables to “absence”. (A–E) indicate the result of grouping of host genera based on the similarity of biogeographical affiliation (see Figure 3). 218 KITADE

Fig. 2. Results of UPGMA clustering of host genera based on the Fig. 3. Results of UPGMA clustering of host genera based on the similarity of symbiont composition using the Jaccard’s coeffi- similarity of biogeographical distribution using the correlation cient as a similarity index. Symbols and numerals (see Table 1) coefficient (r) as a similarity index. Symbols and numerals (see indicate host families and genera, respectively. Capital letters (A– Table 1) indicate host families and genera, respectively. E) indicate the result of grouping of host genera based on the similarity of biogeographical affiliation (see Figure 3). Cluster Analysis based on the similarities of host distribution the Serritermitidae. The genera Reticulitermes (Rhinoter- The host genera were divided into six clusters at a mitidae) and Paraneotermes (Kalotermitidae) were, on the similarity level of 0.1 (A–E: Figure 3 and Figure 1). There other hand, linked most closely to the genera Hodotermop- was no clear correspondence between groupings based on sis (Termopsidae) and Anacanthotermes (Hodotermitidae), geographical distribution or on symbiont composition. respectively. The cockroach genus Cryptocercus, the sole Type III Quantification Method member of the family Cryotocercidae, singly constituted the most basal cluster, and the Mastotermes (Mastotermitidae) The first, second and third axes (axes 1–3) corresponded also singly formed cluster II. to eigenvalues of 0.829, 0.758 and 0.680, respectively. An analysis based on Jaccard’s coefficient generated a Sample scores of axis 1 clearly differentiated Mastotermes similar dendrogram supporting (sub)clusters I, II, III, VI, from other termite genera and those of axis 2 separated IVa, IVb, Va and Vb, whereas clusters IV and V collapsed Cryptocercus and Stolotermes from each other as well as (Figure 2). from the other termite genera (Figure 4a). In the scattergram of sample scores on axes 2 and 3, the host genera in the Kalo- termitidae formed a distinct exclusive cluster (Figure 4b). Flagellate Faunae of Termites 219

Discussion In all of the recent studies of phylogenetic relationships among termites and the wood-feeding cockroach Cryptocer- cus spp.5,12,13,18,19,24,25), monophylies of the Isoptera, the wood-feeding cockroach family Cryptocercidae, and termite families Kalotermitidae, Serritermitidae and Termopsidae, and a group RhinotermitidaeSerritermitidae Termitidae are supported or at least not questioned. Two families each consisting of a single genus, the Cryptocercidae and Mastotermitidae, are always suggested to be the first and second most basal clades18,25). In the present analyses, the host genera in a monophyletic family or group tended to have similar symbiont composi- tion. This tendency was particularly remarkable in the fami- ly Kalotermitidae and the clade RhinotermitidaeSerriter- mitidae, both of which are relatively apical clades5,24). Moreover the two basal clades, Cryptocercidae and Mastotermitidae, each consisting of a single genus, had very different symbiont compositions among all host genera examined. These facts strongly support the idea that symbiont faunae principally reflect the host’s phylogenetic relationships. On the other hand, the similarity patterns of symbiont composition did not clearly correspond to that of distribution of host genera, suggesting that frequency of horizontal transfer of the symbionts, if such transfer occurred, have not been frequent. Although Cryptocercus has a peculiar symbiont fauna, the symbiont genera Oxymonas, , Leptospironympha28), Eucomonympha17) and Streblomastix (Kitade, O. unpubl. data) are found in both Cryptocercus and termite genera. If these symbionts are of the common ancestor of termites and Cryptocercus in origin, their differ- entiation at generic level should have took place within a relatively short period after their symbiotic relationship was established. As an alternative possibility, however, Thorne26,27) speculated that symbionts might have Fig. 4. Scattergram of object scores of host genera calculated by the experienced horizontal transfer between Cryptocercus and type III quantification method. Each symbol corresponds to a host termite, through the ingestion of dead or injured bodies of genus. Symbols and accompanying numerals respectively indi- host of different lineage. cate host families and genera (see Table 1). a, scattergram of the It is intriguing that the genus Reticulitermes had a unique object scores on the first and second axes. b, scattergram of the object scores on the second and third axes. symbiont fauna within the rhinotermitid hosts and its sym- biont fauna resembles that of Hodotermopsis (Termop- sidae). Species of both genera are distributed mainly in the The genera in the Rhinotermitidae and Serritermitidae were temperate regions of East Asia, and they occasionally in- also plotted close to each other. The termopsid and the habit the same logs. The most plausible explanation of such hodotermitid genera also tended to be plotted close together, anomalistic fauna is that horizontal transfer of symbiotic but they did not form distinct clusters as did the kalotermitid protists took place between their ancestral lineages16). genera. 220 KITADE

13) Kambhampati, S. and P. Eggleton. 2000. Taxonomy and phyloge- Acknowledgments ny of termites, p. 1–23. In T. Abe, D.E. Bignel and M. Higashi (ed.), Termites: Evolution, , Symbioses, Ecology. Cluver I thank J. Kojima and A. C. Messer for a critical reading Academic Publishers, Dordrecht. of an earlier draft and T. Matsumoto, K. Katoh and two 14) Kirby, H. 1937. Host-Parasite relations in the distribution of anonymous reviewers for valuable suggestions. This work protozoa in termites. Univ. Calif. Publ. Zool. 41: 189–212. was supported by Grants-in-Aid for Scientific Research 15) Kirby, H. 1949. Systematic differentiation and evolution of flagellates in termites. Rev. Soc. Mex. Hist. Nat. 10: 57–79. (Nos. 14405013 and 14596001) from the Japan Society for 16) Kitade, O. and T. Matsumoto. 1998. Characteristics of the symbi- the Promotion of Science. otic flagellate composition within the termite family Rhinoter- mitidae (Isoptera). Symbiosis 25: 271–278. 17) Kitade, O., T. Maeyama and T. Matsumoto. 1997. Establishment References of symbiotic flagellate fauna of Hodotermopsis (Isoptera: Termopsidae). Sociobiology 30: 161–167. 1) Ando, K., R. Murakami and I. Yamaoka. 1991. On intestinal 18) Lo, N., G. Tokuda, H. Watanabe, H. Rose, M. Slaytor, K. symbiotic protozoa of (Kolbe). Maekawa and C. Bandi. 2000. Evidence from multiple gene Yamaguchi Seibutu 18: 78–79. (In Japanese.) sequences indicates that termites evolved from wood-feeding 2) Cancello, E.M. and O. DeSouza. 1998. A reappraisal of Glosso- cockroaches. Curr. Biol. 10: 801–804. status based on the morphology of all castes with transfer 19) Lo, N., O. Kitade, T. Miura, R.E. Eldridge, R. Constantino and T. from Rhinotermitidae to Serritermitidae, p. 95. In M.P. Schwartz Matsumoto. in press. Molecular phylogeny of the Rhinotermiti- and K. Hogendoorn (ed.), 13th Congress of the International dae. Insectes Sociaux. Union for the Study of Social Insects. Adelaide. 20) Mukherjee, P. and P.K. Maiti. 1989. Description of two new spe- 3) Casgrain, P. and P. Legendre. 2001. The R package for Multivari- cies of flagellates of the genus Dinenympha Leidy (Mastigopho- ate and Spatial Analysis, version 4.0d5—User’s Mannual. Dè- ra: Polymastigida) from Reticulitermes tirapi Chhotani and Das partment de sciences biologiques, Universitè de Montrèal. (Isoptera). Arch. Protistenkd. 137: 91–96. Available on the WWWeb site . Oxymonad from the lower termite Kalotermes sinaicus. Arch. 4) Cleveland, L.R., S.R. Hall, E.P. Sanders and J. Coller. 1934. The Protistenkd. 144: 373–382. wood-feeding roach Cryptocercus, its protozoa, and the symbio- 22) Radek, R. and K. Hausmann. 1994. Placojoenia sinaica n. g., n. sis between protozoa and roach. Mem. Amer. Acad. Art. Sci. 17: sp., a symiotic flagellate from the termite Kalotermes sinaicus. 185–342. Eur. J. Protistol. 30: 25–37. 5) Donovan, S.E., D.T. Jones, W.A. Sands and P. Eggleton. 2000. 23) Sneath, P.H.A. and R.R. Sokal. 1973. Numerical Taxonomy. The morphological phylogenetics of termites (Isoptera). Biol. J. Freenam, San Francisco, CA. Linn. Soc. 70: 467–513. 24) Thompson, G.J., O. Kitade, N. Lo and R.H. Crozier. 2000a. 6) Grassè, P.P. ed. 1952. Traitè de Zoologie. vol. 1, fasc. 1. Masson Phylogenetic evidence for a single, ancestral origin of a ‘true’ & Cie., Paris. (In French.) worker caste in termites. J. Evol. Biol. 13: 869–881. 7) Hausmann, K. and N. Hülsmann. 1996. Protozoology. Thieme 25) Thompson, G.J., M. Lenz, L.R. Miller and R.H. Crozier. 2000b. Medical Publishers, Stuttgart. Phylogenetic analysis and trait evolution in Australian lineages of 8) Hayashi, C. 1956. Quantification theory and its application (II). dry wood termites (Isoptera: Kalotermitidae). Mol. Phylogenet. Ann. Inst. Statist. Math. 4: 19–30. (In Japanese.) Evol. 17: 419–429. 9) Honigberg, B.M. 1970. Protozoa associated with termites and 26) Thorne, B.L. 1990. A case for ancestral transfer of symbionts their role in digestion, p. 1–36. In K. Krishna and F.M. Weesner between cockroaches and termites. Proc. Roy. Soc. London, B, (ed.), The Biology of Termites. vol. II. Academic Press, New 241: 37–41. York, NY. 27) Thorne, B.L. 1991. Ancestral transfer of symbionts between 10) Inoue, T., O. Kitade, Y. Yoshimura and I. Yamaoka. 2000. cockroaches and termites: an alternative hypothesis. Proc. Roy. Symbiotic associations with protists, p. 275–288. In T. Abe, D.E. Soc. London, B, 246: 191–195. Bignel and M. Higashi (ed.), Termites: Evolution, Sociality, 28) Yamin, M.A. 1979. Flagellates of the orders Trichomonadida Symbioses, Ecology. Cluver Academic Publishers, Dordrecht. Kirby, Oxymonadida Grasse, and Hypermastigida Grassi & Foa 11) Jaccard, P. 1902. Gezetze der Pflanzenvertheilung in der Alpinen reported from lower termites (Isoptera families Mastotermitidae, Region. Flora 90: 349–377. Kalotermitidae, Hodotermitidae, Termopsidae, Rhinotermitidae, 12) Kambhampati, S., P. Luykx and C.A. Nalepa. 1996. Phylogenetic and Serritermitidae) and from the wood-feeding roach Cryptocer- relationships among termite families based on DNA sequence of cus (Dictioptera: Cryptocercidae). Sociobiology 4: 1–119. mitochondrial 16S rRNA gene. Insect Mol. Biol. 5: 229–238.