Phylogeny of the Attine Ant Fungi Based on Analysis of Small Subuni' Ribosomal RNA Gene Sequences Gregory Hinkle, James K. Wetterer, Ted R. Schultz, Mitchell L. Sogin* Complete 16S-lke ribosomal RNA coding regions were obtained from the fungal symbiont Of five genera of attine leaf-cutting) ants and two free-living fungi. Phylogenetic analyses with distance matrix, maximum likelihood, and parsimony methods revealed that the attine fungal symbionts are homobasidiomycetes in the order Agaricales. Comparison of the topology of the attine fungal symbiont phylogenetic tree with a tree based on attine ‘ant morphology revealed a congruent branching pattern of the more derived attine ants ‘and their fungal symbionts. The parallel branching pattern suggests a long-term coevo- lution of derived leaf-cutting attine ants and their fungal symbionts. Actines (tribe Attini) are unusual among fants in growing symbiotie fungi for food (1, 2). Despite widespread awareness of attine ants and theie capacity for defoliation, little is known about the evolution of the fungal symbionts responsible for leaf decomposi- tion (3). Unanswered questions include the following: Do different artine lineages cul- tivate the same fungal symbiont? What free-living fungi share most recent common, ancestry with attine fungal symbionts? Did the symbiosis arise multiple times, or was there a single event followed by coevolu- tion? Morphological data were used 10 re solve the phylogeny of attine ants (4). In contrast, the attine fungi possess few mor- phological characters useful in phylogenetic analysis. Here we examine the complete small subunit ribosomal RNA (16S-like FRNA) coding regions (18101821. base pairs) from the fungal symbiont cultivated by five different genera of attines. Our phy- logenetic analyses indicate the attine fungi are homobasidiomycetes inthe order Agari- ales (gilled mushrooms) and that the most derived attine fungi have speciated in par- allel with their attendant ans “The attine ants’ mutuals with fungi has allowed the ants to exploit @ range of food resources not otherwise available (2). ‘The approximately 200 species of attines use a wide variety of material as substrate for growing fungus, for example, insect ass and dry and fresh vegetation (1, 2). The hese known attine ants are the “lea-cut= ting” ants (Ata spp. and Aeromyrmex spp). Wilson (5) considered the adaptation in leaf-euting ants that allowed “efficient uti- lization of almost all forms of fresh vegets- tion” so unusual and successful "tat it can bbe properly called one of the major break- throughs in animal evolution.” “There has been much disagreement con- ceming the phylogenetic affinities of the atine fungi. Under natural conditions, at- tine fungi rarely, if ever, produce sporo- carps, the structures traditionally used in taxonomic determinations of higher fungi Instead, new fungal garens are propagated asexually from a fragment of the fungus garden caried by each new queen from her natal colony (1). On rare, occasions, some atine fungi produce sporocarps in. culture. Mycologists generally assign these sporo- camps 10 one or more genera within the hhomotasiiomycete order Agaricales (6, 7). Some have suggested that the attine fungi axe polyphyletic group made-up of both ascomycotines and basidiomycotines (6,8). ‘Ata copaitae fungal symbiont ————_ Tratymymox brio! Tochangertoo gal mbt ei ‘Srcommex onda tung eymbiont Cyphamyrmexrnosus tenga eben Aoterosigma cota ‘engl symbint Are bombacis ‘Schon commune copes reraue Xrocomus yeni ‘Thanatphors pate Auvicaara peice Maki ita Fig. 1, Distance matrix toe for homobasidomyoste fungi based on 16S-Lke FRNA gone sequences. The tree was infeed from the comparison of pos ‘ions that can be unambiguously algned inal reported fu-legth homobas- somycata 165-ke rRNA gone sequences, his tree, distance matric meth- SCIENCE * VOL.265 + 9 DECEMBER 1994 sed as tha outgroup, ‘ds (15) ware used to infer etatonships forthe fungal symbionts of line ants ‘and representatives of dyerse asiclomycete Ineages. Maka frigia was ‘The aligned data set and the postions used in the ‘analyses are avaiable upon request from the author. 1605,Others have proposed that the artine fungi are all closely related or form a cohesive soup (9). Because of taxonomic uncertai ty, many researchers (9, 10) refer to a strains of attine fungi collectively as Attar ‘myces bromatificus and place them in the polyphyletic Fungi Imperfect. We examined full-length sequences of 16S-like tRNA coding regions from the fungal symbionts of five different species cof attine ants, Cyphomyrmex rimosus, Ap- terostigma collar, Sericomyrmex_bondati, Trachymyrmes bugnioni, and Ata cepha- lots, as well as two free-living Agaricals, Agaricus bisporus and Lepiota procera (11). The attine fungi we chose to study are morphologically and ecologically dissimi- lar and seemed least likely 10 form a ‘monophyletic group. Both T. bugnion’ and S. bondari have “typical” attine fungus gardens that grow as a spongy mycelial sass. Aperostigma collare cultivates fan- us with a distinctive veil, unique to this senus. The fungus cultivated by C. rimosus rows as balls of unicellular yeast. Finally, the fungus gardens of the leaf-cutting ant A. cephalotes are morphologically similar to those of Trachymsrmex and. Serico- ‘myrmex but grow faster and can support colonies of up to several million workers compared with 100 to 5000 workers for colonies of the other attine genera Comparison ofthe attine fungal symbi- cont 16S-like RNA genes clearly demon- strates that four of the five taxa sampled represent distinct basidiomyeete genera (12). Only. Sericomyrmex and Trachy- smyrmex are sufficiently similar (two nucle- otide changes) to permit assignment with- in the same genus. Pairwise comparisons between the other attine fungal 16S-ike ARNAs range from 26 changes for symbi- fonts of Ava and Cyphomyrmex to 80 changes between Apierostgma and Serico- ‘mymmex symbionts. This number of differ ences is similar to that observed between fungal families, for example, 35 nucleotide differences between the 16S-like ¢RNAS of Agaricus bisporus and Lepioza procera. In contrast, the similarity of Tzackymgrmex and Sericomyrmex fungal symbionts sug- gests a very recent divergence. “To asses the origin of attine ant fungal symbionts, we inferred phylogenetic trees using parsimony (13), maximum lkelthood (14), and distance matrix (15) analyses of 16S-like #RNA sequences. All phylogenetic reconstructions show the attine fungal sym= bbonts to be members of the order Agari Heke ra Soa, Maine Baga aboatay, Woods Hola, MAGz54s, USA SC tera, Museum of mparave Zelogy, Ha ‘ard Unkeraty, Cambio, MA‘02138, USA TER Shut Boparment of nlomaogy, Comell nie: syste, NY 14853, USA "To whom corecpondence shod be saAese, 1698 cales (Fig. 1). The fungal symbionts of “higher” attine ants (Atta, Cybhomyrmex, Sericomyrmex, and Trachymyrmex) represent a wellsesolved, coherent phylogenetic sroup that diverged after the separation of the fungal symbiont of Apterostigma. More importantly the evolutionary relationships of the higher attine ants and their fungal symbionts are entirely congruent (Fig. 2) Parallel patterns suggest cospeciation and codivergence among higher attine fungi and their associated ants (12, 16). Ale though we are not able to resolve the branching order of free-living Agaricus, Lepiot, and the fungal symbiont of Apuerostgma (17), the congruency of the ant and fungal tres is evidence of stable coevolution for millions of years (18). The ant tribe Attini is thought to have ‘originated about 50 million years ago (19). The divergence between Caphomyrmex and Trachymyrmex is at least 25 to 40 million years old, as evidenced by the occurrence of these two genera in the Dominican Amber (20). The genetic divergence among the fungi of the higher attines (Fig. 1) is eon- sistent with a mutuaism of considerable antiquity Artine fungi that may not have repro- duced sexually for millions of years occa sionally produce in culture agarie-like spo- rocatps. The sporocatps from the fungi of different genera of attines are viewally iden- tical morphologically (7). The lack of phe- rnotypic divergence in sporocarps from taxa Fungi ptyvogeny that diverged >50 million years ago is strik- ing and suggests that sporocarps may indeed be produced in nature, albeit rarely, or that the genes responsible for sporocarp forma- tion are expressed for altemative functions and hence retained, ‘Attine ants and the attine fungi are mutually dependent (21). The fungus re- lies on the ants, not only for substrate and for protection from competing microor- ganisms but also for propagation. In turn, the fungus is an essential part of the ants! diet. This study suggests that the diversi- fication of the higher attine ants and fungi took place within the symbiosis, and therefore, well-developed symbiont-spe- cific adaptations should be expected. The cco-adaptation that led ancestral leaf-cut= ting ants to use fresh vegetation may have involved a physiological breakthrough on the part of the fungus (10). The yeast of C. rimosus evolved from a mycelial ances- tor under the care of an ancestral Cypho- myrmex species. The adaptive significance of this change in fungal growth form is unknown, ‘The coevolution of attine ants and their fangi is an excellent example of ov sym- bionts, often unseen or overlooked micro- organisms, can underlie the preservation of more obvious and much studied macro- biota. In this example, it is unlikely that any of these attine ants would survive if their fungal partners were exterminated by some environmental insult Fig. 2. Phyiogenstic ree ‘Ant phylogeny of symbiotic attine ants ees and fungi. Theatine fun ‘al trea was inlared by aaa Comparison of unambig | _Sehizophytum commune tous aligned postions Of smal suburst rRNA [—— Asaricu sporus sequences with max ‘hota bombacina mum parsimony (13) cone “The postions usedin the analysis are identical to sol ‘Atta cophatotes those used in ig 4. The percentage of 100 boat. ‘Sericomyrmex bondart eo tewrglngs end L subsequent heuristic s ‘searches that support 9 cnn ‘dertica clades above the 50% loves incteat- Cyphomyrmex rimosus ee ere attne ant phylogeny is oa Apterostigma collare erived from a cladistic analysis of 44 prepupal sof Boletus santas ‘worker larva mercholog ‘eal characters trom 51 erocomus chysontoron altine spacies (4). The topology of the core ‘Tanetephonss pratcolo sponding higher ating ant and fungal symbi- Auriulana polycha fonts are deinaatod with athek ina, Mra tigi Blophavcatabaslienss SCIENCE + VOL.266 * 9 DECEMBER 199412, 1%. REFERENCES AND NOTES INA. Wicbr, Mam. Ar Pio, Soc 92,1 (1872, B, Holler and EO. Wison, To Arts Marverd Ur Pres, Cambridge, MA, 1900 Water, InNeurematand Eves insect ‘ecient ana A Napa, Es, (West ‘ow, Bou, 00,190), go. 308-228, TOR. Schutz ard FL Mor Sjstom. Entomol, in ess, 0. Wan nF Arts and Laat Outing Arts, C.8. Logon aed PR. . Vander Mer, Eds. (Weston, Boulder, 00, 198). 13 NLA obo” Rew. tome. 8, 154 (1638); Senco 189,557 (196) n Pst Fungus Sys, LF Bara, 6 (Hated, Mert, Ns, 1079, pp. 77— ‘A Hervey, CT. Rogerson. Leon, Briton 28, Bes orth ‘A Kamat, M. Decharma, 6, Feta, In (2 231-246 1.4. Sraing and Ru Powel, Exparents 42,962 {1086 Chara J Pow. J. Stadio. Ihesct Fungus inact, NW. Ne Cali, PM, Hammond, iF. Webber, Eas. (Acadame Prose, Now Yor, 1969, 1.89120. Rou Powel and D. J. Sradirg, Tans. 8. Mot So. 87,205 (1980); H. Kel in Inox of Pray (CAB.CML, Kaw, Sury, UK, 1075) ot 4, pat ‘ere ate hangs cuties were supple by J azn fambiont feu and, Chapel ‘orto Cyprammex inosus,Acterostisra 0 lee, Sercarmrmax bond. and Ata exphattes Preparaten of genome ONA (. Lae and J “ayer: n POR Protocols: A Gude fo Mets and ‘Agpiatons, M.A. ri, D-H, Gating, dS ST Whe, Eds (acoder Poss, Sn Cod, CCA, 1800, pp. 282-287 andthe pohmerase chan reaton (POR artcaton Mean H, J Ewood, Ske Stoke ML Soin, Gare 7,491 (1980 and Sequencing of 16S YANK Pa Enos, (ben, MCL. Sag, Mol Bot vl 2, 500 (1085) wre prove desctbod. GenBank accession Fubar area ols: Aterostna cole hal ‘SymbiertL0ese5), Aa cophabts gal mbit ‘hsb. Qeronvmat ars ava Sree: {U02500), Soneamrmer donden nga symbiont {Uaa590), acryyrmox Boge gal sybion (onss7),Agareue Depo 30058), and Lope rosea 36869). Nolet dlerenosebetwoan 168-he NAS of the atine fungal symbionts ro contr wi 3 [rvogerate andjoce based on paral 255 ho [Rk sequsnoa deterinaters worn genera core. Sprang to those nuded nthe sty [A Gri A Paton TRS, U. GN, ‘Soanco 26,1001 (1584) 1. Sword, PAUP. Prjgonetic Anas Usb Parsrary, arson 31.1 pgs NatralHetery Su ‘wy, Chapa, 1850) 6G, sa, H. Mats Hagstrom, Overbok, (Cem Apo Bisa 10,4 (1908, Gi. Oban, Metrods Emre 164, 7081988). (. Mtr ad DR. Exec, Coovliton,D. J Futoma and M, Stk, Es, Srey, Sundog MA T0805, Macraum parsimony angie (wth branch and band seach stages reduced wo tres of 425 Steps. Boos conta th hese fuga Bont anajzod na coherent pyogenes ast age. Honaver. becstap parumory (4) max ‘mum ikenood 15, and dane mati (76 mat (08 donot ies prefered branching ode Semong Agar, Lepit, ant tagal smblon ot ‘esas sina tha bean possible te demonstrat thatthe fungal ‘5mionto Ateostigms branched bere te d ‘erence of Lops and Agaricus, two coneetng Iarretatons wou havehad thane been consi (ean tho trot wo Independant org fsb ‘te tunat wih atnearts ay have cose. Fo ‘tematie terretaton,Lepota and Agaricus may Faw ezeapad the symbioesin avr ofa eng Hee. 10. A Weber, Pc nt. Cong. Eta. 2, 459 (958. 20, ©. Baro Uta, Sut. Ba Nata, Soe 8 58,1 (ia; E 0, Witor, Soince 228, 205 (185), 21. MLM Marten ivrstate Merona reactors Ingostes Fungal Braye Artiropod Baba). M. IM Marie, €2 (Camoce Macs, NY, 1987, 91-125, 22, We thank J. Cazn and |. Chapel forthe tine fungus cultures; M. Wetter, |. Chapel, ©. Hb batt D. Mile, and Yu fr coriments onthe aruscrt David Bemudos for samplos Lap? taproot: NSF, fe Natlona Geograph Scaely KW). the G. Unger Veleson Foundation, and NN ant number Ga29eH fo MLS} fe fea al suppor. 19 June 1994; acoapted 21 October 1094 Naturally Occurring Variation in Bristle Number and DNA Polymorphisms at the scabrous Locus of Drosophila melanogaster Chaogiang Lai,” Richard F. Lyman, Anthony D. Long, Charles H. Langley, Trudy F.C. Mackay ‘The association between quantitative genetic variation in bristle number and molecular variation at a candidate neurogenic locus, scabrous, was examined in Drosophila mela: rnogaster. Approximately 32 percent of the genetic variation in abdominal bristle number (21 percent for stemopleural bristle number) among 47 second chromosomes from a. ‘atural population was correlated with DNA sequence polymorphisms at this locus. ‘Several polymorphic sites associated with large phenotypic effects occurred at interme diate frequency. Quantitative genetic variation in natural populations caused by alleles that have large effects at a few loci and that segregate at intermediate frequencies Ccontlicts wth the classical infinitesimal model of the genetic basis of quantitative variation. Konowledge of the genetic basis of quanti- tative characters is important with regard to edicine, the improvement of domestic species, and our understanding of evolution, yet litde is known about the particular Mendelian variants that give rise to the heritable component of these traits. One hypothesis is that allele variation at loci important in the development of a particu- Jar trait is a major source of quantitative differences in that trait (1). The numbers of abdomninal and sternopleural bristles of Dro- sophia melanogaster are typical quantitative characters (2). Because Drosophila bristles are sensilla (sensory organs) of the periph- eral nervous system, candidate genes for bristle number teaits are the 10 to 20 pro- neural and neurogenic loei that determine the presence or absence of sensory hairs (3) Alleles of large effect at some of these loci tay contebute to the response to artificial selection for high and low bristle numbers, (4), and insertional polymorphisms in the proneural achaete-scute complex (ASC) are associated with naturally occurring genetic variation in bristle number (1). The sca- brows (sca) locus encodes a signal protein ‘Crlaland Landay, Canaries Poplaion aaa ‘Unorsty of Ciloma t Dav, Dav, CA So6t6, USA RE Lyman ard F.C Mackay, Department of Geet les, Noth Carne State Unters, Flog, NC 27035 — 7614, USA Lang, Contr er Poputon Belogy, Urey of Calin af Da, Dave, CA 95616, USA, ard Depa ‘pert of Belay, MeMaster Unis, Haiton, Orta, Canad as "To wham conespondsnce shold be sairesod, SCIENCE + VOL. 266 + 9 DECEMBER 1994 {important in lateral inhibition ofthe devel- ‘oping nervous system, and mutant sca al- Teles have large effects on bristle number and eye morphology (5). We have now tested the hypothesis that allelic variation at sca contributes to quantitative genetic, variation in natural populations of D; mela- nogester by associating molecular polymor phisms at this locus with genetic variation in bristle number. We determined the mean abdominal and stemopleural bristle numbers for each, sex from 47 independent second chromo- some lines extracted from a natural popula- tion and placed in an isogenic genetic hack- ground (6) (Table 1). The overall means (SE) were 1630 * 0.42 abdominal bris- tes and 16.15 * 0.22 stemopleural bristles. ‘Assuming additivity, the total additive ge- netic variance (9) of the second chromo- some was estimated as 1.23 for abdominal bristle number and 0.64 for sternopleural bristle number (7). The additive genetic covariance between both characters was 039, with a genetic correlation coefficient cof 0.43 (7). These estimates are consistent with previous observations (8). Restriction map variation of a 45-kb region including the sea locus was quanti fied among the 47 chromosomes (Fig. 1 and Table 1). There were 18 restriction site polymorphisms and 25 length poly- morphisms (insertions and deletions). Sin- flestranded conformation polymorphism (SSCP) was determined (9) for three frag- rents that encompass the last intron and 1697