Origin of a Complex Key Innovation in an Obligate -Plant Mutualism Author(s): Olle Pellmyr and Harald W. Krenn Source: Proceedings of the National Academy of Sciences of the United States of America, Vol. 99, No. 8 (Apr. 16, 2002), pp. 5498-5502 Published by: National Academy of Sciences Stable URL: http://www.jstor.org/stable/3058521 Accessed: 24/01/2010 21:31

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=nas.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

National Academy of Sciences is collaborating with JSTOR to digitize, preserve and extend access to Proceedings of the National Academy of Sciences of the United States of America.

http://www.jstor.org Origin of a complex key inno vation in an obligate insect-plant mutualism

Olle Pellmyr*t and Harald W. Krenn*

*Department of Biology, Vanderbilt University, Box 1812 Station B, Nashville, TN 37 235; and *Department of Evolutionary Biology, Institute of Zoology, University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria

Edited by May R. Berenbaum, University of Illinois at Urbana-Champaign, Urbana, IL, and approved January 30, 2002 (received for review November 2, 2001) Evolutionary key innovations give organisms access to new eco- cle s to propose a possible developmental genetic basis for the logical resources and cause rapid, sometimes spectacular adaptive trait. radiation.The well known obligate pollination mutualism between and is a major model system for studies of The Functionof the Tentacles.The pollinating yucca genera coevolution, and it relies on the key innovation in the moths of Te geticula and Parategeticula constitute a monophyletic group complex tentacles used for pollen collecting and active pollination. wi :hin the (Fig. 1). Jointly they contain at least 25 These structures lack apparent homology in other , making ex :ant species (5), two of which are derived nonpollinating them a rareexample of a novel limb. We performed anatomical and Tegeticula species that oviposit into yucca fruit created by behavioral studies to determine their origin and found evidence of co existing pollinator species (16). The sister group Prodoxus a remarkably simple mechanism. Morphological analyses of the co exists with the pollinators on yuccas but feed as larvae on plant tentacles and adjacent mouthparts in pollinators and closely re- pa rts other than the seeds. Their radiation was thus directly lated taxa showed that the tentacle appears abruptly in female fa?:ilitated by the pollinator radiation. Together these genera pollinating yucca moths. Several morphological synapomorphies co nstitute a major adaptive radiation on yuccas, with a species between the galeae, which constitute the characteristic lepidop- di),ersity more than 20-fold that of their sister group, the teran proboscis, and the tentacle suggest that the tentacle evolved nc npollinating seed-parasitic Mesepiola, whose larvae feed on quicklythrough expression of the genetic template for the galea at ph ints in the Nolinaceae (17). an apical growth bud on the first segment of the maxillary palp. Female yucca moths possess unique tentacles on their mouth- Behavioral data indicate that tentacle and proboscis movements pa rts that are used to actively pollinate host flowers where they are controlled by a shared hydraulicextension mechanism, thus no ov iposit. The female moth gathers the glutinous pollen of yucca new mechanism was needed for tentacle function. Known devel- flc 'wers by scraping it off the anthers with her tentacles. The opmental paths from other insects can explain the origin of this po Ilen is immediately compacted by using tentacles and some- sex-specific key innovation in a few steps. tir les the forelegs as well, and placed as a solid batch on the co ncave posterioventral surface of the head (Fig. 2). The pollen iss and to 10% of the mutualisms between plants and pollinators provide m; may approach 10,000 grains weigh up O bligate )th mass Prolific maintains batch some of the most examples of coevolution (1, 2). ml body (18). pollen coating apparent co and the tentacles are not involved in its retention. A association of this kind, between yucca moths hesion, long-recognized ter the moth seeks out (Prodoxidae) and yucca plants (Agavaceae), has become an Ai pollen gathering, flowering yucca model in how mutualisms p1 ints where she oviposits into (Tegeticula) or near (Paratege- important understanding obligate As is the female flexes her coevolve In this association, established at least 40 million tic ula) pistils. oviposition completed, (3-6). itacles and uses the to remove a small are moths, te] apical portion pollen years ago (7), yuccas pollinated exclusively by yucca floral and whose larvae in turn consume some of the lo id from her batch. She walks to the stigma very developing yucca the on it. In all but one host seeds. This has been an and e liberately places pollen species, evolutionarily ecologically highly - line the interior of the hollow and the successful association, with some 30-45 stigmatic papillae style, yucca species (8, 9) )th in the with 10-20 motions much of the m' packs pollen repeated bobbing being important vegetation components throughout the course of 3-10 sec which is available as North American deserts and semiarid n (Movie 1, regions (10). information on the PNAS web In Prior of the coevolution between moths and su pporting site, www.pnas.org). analyses yucca - the host has a have shown that the transition from to single exception, (Hesperoyucca whipplei) yuccas antagonism ca and the moth the same mutualism involved in Cap-shaped stigma, pollinates by using primarily quantitative changes already behavior on the as is used for collection rather than novelties. The one agging stigma pollen existing traits, evolutionary the anthers. exception is the evolution of elaborate tentacular mouthparts in on the yucca moths, used for handling pollen with great precision. M rterials and Methods These tentacles are an evolutionary key innovation, both in the ale and female moths of six species were included in the study sense that it is a truly novel trait that evolved quickly (7) (Fig. 1) ig. 1). Two outgroup taxa were used to determine the basal and that it is linked to an adaptive radiation (11-13). Under- ( COndition. Nemophora degeerella(Adelidae) is a basal member of standing how this trait evolved, then, is central to understanding th superfamily Incurvarioidea, which includes Prodoxidae. the coevolutionary history of diversification and changing inter- p odoxus decipiens represents the pollinator sister group, thus actions between yuccas and yucca moths. Although reported be ing a close relative of the common ancestor of the tentacle- when the relationship was first described over a century ago (14, be aring moths of Tegeticula and Parategeticula. A clade of three 15), no analyses have been performed of tentacle anatomy or homology. Here we present anatomical data from phylogenetically piv- This paper was submitted directly(Track II) to the PNASoffice. otal moth species indicating that this complex key morphological tTcwhom reprintrequests should be addressed.E-mail: [email protected]. for the mutualism has a We also Th trait surprisingly simple origin. publicationcosts of this articlewere defrayed in part by page charge payment. This use trait expression in the pollinators, their nonpollinating sister artcle must therefore be hereby marked "advertisement"in accordancewith 18 U.S.C. group, and derived species that have secondarily lost the tenta- ?1;34 solely to indicatethis fact.

5498-5502 | PNAS | April16, 2002 | vol. 99 I no. 8 www.pnas.org/cgi/doi/10.1 073/pnas.072588699 CS3

: ( cr u, Pr

czr

X 7Tegeticula /

tentacles lost

.'. tentacles present (pollination) Prodoxidae Adelidae yuccas colonized

Fig.1. Phylogeneticpositions of the sixspecies used in the study.Names of speciesused are given at top. Triangle width reflects species richness for the three yucca-feeding genera and their sister group. Spacing between prodoxid genera an i the Tegeticula species triad is proportional to time, and the bottoms of triangles give deepest known radiationswithin each . The deepest split (Mesepiola vs. others) is estimated to 44.1 + 10.6 millionyears ago. The internode from the Prodoxus-pollinatorgenera split to the pollinator genera split, along whic 1 the tentacle evolved, is so short that estimated ages of the three genera overlap, thus the uniform tentacle seen in all pollinator moths must have evolved veryquickly. Data are fromrefs. 5 and 7, and O.P.and M. Balcazar-Lara, unpublished data.

Tegeticulaspecies was used, includingthe pollinatorsTegeticula pc llinating species of Tegeticula and Parategeticula (Fig. 3). yuccasellaand Tegeticulacassandra, and the derivednonpollina- M ales possess a small cuticle elevation in the same area (Fig. 3B). tor Tegeticulaintermedia. T. intermediarecently diverged from T. N() traces of a tentacle were found in either Prodoxus or cassandra (5), thus providing information on loss of the tenta- NMmophora. No variation in tentacle morphology was found cles. A single representative of the smaller pollinator genus an the three The was included to long analyzed pollinator species. nonsegmented Parategeticula,Parategeticula elephantipella, te itacle is 2.4 + 0.23 mm-long (n = 10) in T. yuccasella, about cover the phylogenetic range of species with tentacles. Sample 10 ,Lmin cross section at and it toward the data are in Table 1. O midpoint, tapers tip. provided Di iring rest, the tentacle is coiled under the head when the Live moths of all Tegeticula, Prodoxus, and Nemophora sam- fe nale does not carry pollen (Fig. 3) and is around the pies were placed for 24 h in alcoholic Bouins' fixative and then wrapped lien load when one is The surface transferred to in 70% ethanol. For Paratege- PC present (Fig. 2). (Fig. 3C) permanent storage covered with and has numer- ticula, internal anatomy was analyzed in a female moth stored is annulated, densely microtrichia, s trichoid sensilla with a hooked These sensilla are located directly in 70% ethanol, and two dried male and female indi- Ol tip. viduals were examined. Studies of external anatomy were per- pr imarilyon ventral and lateral sides but appear on all sides near formed on dehydrated and gold-coated heads by using scanning th apex. electron microscopy. Internal anatomy was examined by using The most prominent internal feature of the tentacle of polli- serial semithin sections of specimens embedded in epoxy resin. na tors (Fig. 4) is longitudinal musculature consisting of several : Tentacle size in T. intermediawas compared by scoring relative hundred minute muscles distributed mostly along the ventrolat- length on a linear scale from 0 to 10, where 0 was no trace and er il walls, although some fibers have an oblique dorsolateral 10 representedthe longest rudiment,which was one-halfof the co urse (Fig. 4B). These muscles permit recoiling and possibly in T. length the sister species cassandra. Ten fixed individuals of so me degree of lateral movement of the tentacle. They are quite each sex were used for the comparison. di;;tinct from muscles attached to the base of the second max- Movement in and patterns tentacles, palps, proboscis during ill iry palp segment in all studied species, where they serve as pollen collection, pollination, and drinking were observed to fl :xors and extensors. Other features of the tentacle include a determine whether extension mechanisms are shared between ckened dorsal and and a tentacles and other Fourteen female T and T epidermis cuticle, nerves, single parts. yuccasella lo trachea cassandra in the field were individually in igitudinal (Fig. 4). gathered placed glass In the derived T. tentacle rudiments vials with a Yucca filamentosa flower. As moths became active at nonpollinator intermedia, e in both sexes and there is no late dusk, we observed mouthpart movements through a 10 ar present (Fig. 4B), significant in > lens under red narrow-bandwidth illumination (peak 626 nm) di 'ference length between them (t test, P 0.12). Micro- chiation and a reduced number of hooked sensilla are still esent (Fig. 5). A reduced number of muscles are present Results re ;ardless of length of the rudiment (Fig. 5B). Tentacles are thus Tentacle Anatomy. The tentacle emerges frontally on the first se cually monomorphic in the nonpollinators but dimorphic in maxillary palp segment and is found only in females of the th eir pollinating ancestor.

Pellmyr and Krenn PNAS | April 16, 2002 | vol. 99 I no. 8 | 5499 Fig. 2. Head of female Tegeticula carnerosanellawith yucca pollen load. Blackarrow = left tentacle; white arrow = proboscis.

Comparisonwith the Proboscis.The galea is an appendage that emerges next to the multisegmented maxillary palp from a shared basal maxillary plate. In , the two galeae zip together after adult emergence to form the highly distinctive proboscis that is used for drinking (Fig. 3 B and D). The Fig 3. (A) Mouthparts of female T. yuccasella, lateral view. Tentacle (t) ori from first of tentacle coiled morphological analyses revealed a striking number of shared ginates segment maxillarypalp (mpl); laterally between the tentacle and 3 C and rcm the proboscis(p); distal segments of the maxillarypalp (mp)covered with specializations proboscis (Fig. sca les. Male of T. with a minor betweenthe tentacle and include frc (B) yuccasella maxillarypalp (mp) showing D). Synapomorphies proboscis ntal elevation on first segment instead of a tentacle; proboscis(p) coiled nonarticulationand coiling ability,dorsally thickened epidermis unjer the head. (C) Tentacle surface of T. yuccasella with microtrichiated and cuticle, fluted sensory bristles, and similarlyarranged in- surface and prominent trichoid sensilla (s) with hooked tips. (D) Proboscis trinsiclongitudinal muscles. Modifications in the tentacle from surface of female T. yuccasella with microtrichiatedsurface and trichoid the proboscis include increased thickness and increased number sersilla with straight tips. of muscle fibers, absence of a medial food groove, and hooked tips of the sensilla. No common features with the maxillary palp were found apart from the surface microtrichia. sic,n. During feeding, the probosciswas straightand in line with thl body axis,with the tip touchingthe substrate.Meanwhile the MouthpartMovements. Behavior observations showed that the te] itacles,which are not used duringfeeding, were partlyinflated proboscis was extended whenever the tentacles were, indicating so that they formeda semicirclein femaleswithout pollen loads. a shared hydraulic mechanism for proboscis and tentacle exten- Di iring pollen collection and pollination, tentacles were fully ini 'lated and nearly straight, whereas the nonparticipating pro- boscis was even more inflated thanwhen feeding,being straight Table1. Speciesused and geographicorigins of samplesfor and raisedabove the bodyaxis so thatit did not touchthe ground. the analyses W hen a pollen load was present, both proboscis and tentacles Species Locality we re already partly uncoiled as they wrapped around the pollen. In such circumstances, concerted tentacle movement with the Adelidae pr Tboscis was reduced relative to individuals without a pollen Nemophora degeerella (L.) (2, 2) Austria: Vienna, Satzberg lo id. Regardless of pollen status of a moth, the terminal Prodoxidae ixillarypalp segment also moved forwardand backwardin Prodoxus decipiens Riley (2, 2) U.S.: Texas, Jefferson un ison with tentaclesand proboscis,as predictedfrom a shared P. elephantipella Pellmyr & Mexico: Veracruz, near Huatusco hy draulic mechanism. Balcazar-Lara(0, 1) T. yuccasella (Riley) (4, 4) U.S.: Nashville, TN Di icussion T. cassandra Pellmyr (5, 5) U.S.: Florida, Lake Placid Tt e tentacular appendages of the maxillary palp in the female T. intermedia (Riley) (6, 6) U.S.: Florida, Torreya State Park :ca moths are a rare example of a complex key innovation that Numberof individuals(male, female) used for the anatomicalanalyses are en lerged in one clade without any homologous structure in given in parentheses. re. ated taxa. A considerable number of morphological synapo-

5500 | www.pnas.org/cgi/doi/10.1073/pnas.072588699 Pellmyrand Krenn 'A i . 4",._ -

100pm , B 3 -s

^~ pmE ~ ~4, ~

A~~~ o. mp1-

'100Lm -' 100tm Fig.4. (A) Internalanatomy of the tentacle of female T. yuccasellain tentacle musculature thickened longitudinal section shows longitudinal (tm), Fig 5. (A)Tentacle rudiment (t) on firstmaxillary palp segment (mpi) of dorsalepidermis and cuticle (arrowheads), a nerve (n), and a trachea(tr). (B) fernale T. intermedia; second maxillary palp segment (mp2) with scales;mi- Crosssection through tentacles (left and rightsections) and proboscis,the crctrichiated proboscis (p). (B) Tentacle rudiment of femaleT. intermedia in T. latterof whichis composed of the two galeae(Center), in female yuccasella. orgitudinal section emerging from first maxillary palp segment (mpi) con- Sharedinternal features include proboscis muscles (pm) and tentacle muscles tai nstentacle musculature (tm). (tm)in ventrallumen, thickened dorsal epidermis and cuticle(arrowheads), nerve,and trachea (tr).

flc ttened side of the galea that forms the central food canal of morphies are shared between the tentacle and galea, suggesting th : proboscis is lost. that the genetic template for the galea is fundamental in producing the tentacle as well. They include the rare condition Genetic Basis of the Tentacle.From a morphological perspective, of a nonarticulate and coilable limb, similarly ar- it is striking that such a complex structure as the tentacle has ranged novel musculature, dorsally thickened epidermis and ev olved without any homologous features in the related prodoxid cuticle, and specific sensilla. Both structures arise from the same mi)ths. The simplest explanation for the shared specializations basal maxillary part, the stipes tube, which has been functionally be tween the tentacle and galea is that a shared developmental in interpreted to be a hemolymph pump the higher, glossate pa thway is involved. Although the galea is basal to all insects (25, The behavior data indicate that the Lepidoptera (19, 20). 26 ), this maxillary appendage is highly modified in most Lepi- extension mechanism is shared as as tentacle hydraulic well, dc ptera. The most basal clades (Micropterigoidea, Agathipha- extension led to extension and move- simultaneously proboscis go idea, and Heterobathmioidea) have a minute protrusion, but ment of the terminal The lesser move- maxillary palp segments. th dclade Glossata, which constitutes about 99.9% of all de- ment of tentacles as the is extended for is proboscis drinking 'ibed Lepidoptera, is characterized by elongated and con- for two reasons. First, the volume of the tentacle expected larger cted galeae that form the distinctive proboscis (20, 27). creates slower extension when pressure is adjusted for proboscis W hereas the earliest proboscides lacked intrinsic musculature extension. Second, when the moth carries a pollen load the f r tight coiling, this trait evolved well before the origin of the tentacle is already partly uncoiled without internal liquid pres- odoxidae (28-30). The yucca moths have a relatively short sure as it wraps around the pollen, and influx will cause only oboscis consisting of loosely connected are further extension as it reaches beyond the uncoiling. pr galeae; they on wet surfaces of flowers, Modifications in the tentacle from the galea are modest and sP layed apart yucca allowing liquids ch as water and nectar to be imbibed force mostly quantitative. The tentacle has a greater diameter than the su by capillary (31). Information about is limited but galea, and the number of ventrolateral muscle fibers is much galea developmental genetics increased. The bristle-like sensilla are clustered in the ventral i formative in developing a hypothesis for tentacle development. ie bud of the shows of Distal-less region, are larger, and have gained the distinctive terminal hook. TL growth galea expression (Dll), They aid in handling the pollen and seem to evolve easily among a homeodomain transcription factor characteristic of and re- pollen-collecting insects. Similar hooked-tip sensilla have quired for the development of distal limb structures (32-34) in evolved independently at least 11 times among more than 50 bee a 1)road taxonomic range of insects (35, 36). Galea development species that collect pollen from flowers whose anthers are is usually sexually monomorphic, including in the yucca moths. concealed in narrow tubes (21-24). In bees, the hooked hairs In contrast, functional tentacles are sexually dimorphic in pol- appear on mouthparts or foretarsi, depending on what body part lir ating yucca moths, but sexually monomorphic partial expres- a particular species uses for pollen-gathering. Finally, a medial si( mnof maxillary tentacle rudiments in nonpollinating T. inter-

Pellmyrand Krenn PNAS I April 16,2002 I vol. 99 [ no. 8 I 5501 mediais evidence that the developmentaltemplate is presentin elc)ngated galeae of Lepidoptera. Prior work on the interaction both sexes of pollinators but repressed in males. ha s suggested that the tentacles are the only truly novel mor- A simple explanation for the observed patterns holds that the ph ological trait in these moths, and the present findings show basic tentacle is an appendage using the genetic template for the th;it acquisition of this complex trait and its mechanism of galea. Dll is expressed in all maxillary palp segments (35, 36), and m( Ivementmay have been evolutionarilysimple. Meanwhile, the the tentacle would require protracted expression at an apical bud be havioralcomponent of pollination is likely derived from a on the first palp segment. Limitation to expression in females co nmon probing behavior for nectar in floral tubes in more could be determinedby as little as a single gene, such as bab, ba sal prodoxids(4) and on the wet stigmas of yuccas, thus a known to homeotic and other to cause sexual integrate pathways pa ssive, less efficient pollination mechanism may have preceded in A basis is con- dimorphism Drosophila (37). simple genetic th \ origin of the tentacles and active in the sistent with the reversal to sexual in the pollination yucca rapid monomorphism )ths. Given the significanceof these organismsin studies of derived nonpollinators, but could also be explained by a more with the of evolutionof trait with at least one In the evolutionaryprocesses, coupled rarity complex major-effect gene. contrast, ne w it shouldbe to test the of rudiment in limbs, important explicitpredictions variable degree expression nonpollinators, rang- ab out of that derivefrom this from a blunt to a short tentacle with some intrinsic patterns gene expression hypoth- ing point s about tentacle musculature,indicates slow functionalloss throughmutations in e origin. different of the tentacle This is parts template. hypothesis dedicate this to the of Ebbe who consistent with life reconstruction of the paper memory Schmidt Nielsen, history nonpollinating Colitributed to our of basal evolution of of greatly understanding Lepidop- yucca moths, which suggests loss pollination behavior was tera. We thank Kari and Manuel Balcazar-Lara for to a shift from in flowersto in Segraves help secondary ovipositing ovipositing hering specimens, Ursula Hannappel and Alexander Pernstich for the fruit (5), making the structuresand behaviorsassociated with pri of the semithin sections, and Jim Kane for on bee to loss ab- -paration guidance pollinationredundant and subject gradual through serisilla. Florida State Parks provided permission to gather samples in sence of purifying selection. To rreya State Park. Behavior experiments were performed at Archbold In conclusion, the morphological data strongly indicate that Bii)logical Station. This work was supported by the National Geographic the unique tentacles of female yucca moths originated through So :iety, the National Science Foundation, and the Austria Science Fund expression in a novel site of the genetic template for the (P 13944).

1. Herre, E. A. (1999) in Levels of Selection in Evolution, ed. Keller, L. (Princeton 19. Banziger,H. (1971) Mitt.Schweiz. Entomol. Ges. 43, 225-239. Univ. Press, Princeton), pp. 209-237. 20. Krenn, H. W. (1990) Zoomorphology 110, 105-114. 2. Thompson, J. N. (1999) Science 284, 2116-2118. 21. Miiller,A. (1995)Entomol. Gener. 20, 43-57. 3. Bogler, D. J., Neff, J. L. & Simpson, B. B. (1995) Proc. Natl. Acad. Sci. USA 22. Alves dos Santos,I. & Wittmann,D. (2000) PlantSyst. Evol. 223, 127-137. 92, 6864-6867. 23. Michener,C. D. (2000) The Bees of the World(Johns HopkinsUniv. Press, 4. Pellmyr, O., Thompson, J. N., Brown, J. & Harrison, R. G. (1996)Am. Nat. 148, Baltimore). 827-847. 24. Thorp,R. W. (2000)Plant Syst. Evol. 222, 211-223. 5. Pellmyr, O. & Leebens-Mack, J. (2000) Am. Nat. 156, S62-S76. 25. Snodgrass,R. E. (1935) Principlesof InsectMorphology (McGraw-Hill, New 6. J. L. Ecol. Lett. 277-287. Bronstein, (2001) 4, York). 7. & Leebens-Mack, J. Proc. Natl. Acad. Sci. USA 96, Pellmyr, O. (1999) 26 Chapman, R. F. (1998) The Insects: Structlure and Fulnction (Cambridge Univ. 9178-9183. Press,Cambridge, U.K.). in Intermountain eds. 8. Reveal, J. L. (1977) Flora, Cronquist, A., Holmgren, 27 Kristensen,N. P. & Skalski,A. W. (1998)in Lepidoptera:Moths and Butterflies A. N. J. L. & P. K. York Botanical H., Holmgren, H., Reveal, Holmgren, (New 1. Handbookof Zoology,ed. Kristensen,N. P. (de Gruyter,Berlin), pp. 7-25. Garden New Vol. 527-536. Press, York), 6, pp. 28 Kristensen,N. P. (1968)Ent. Medd.36, 239-293. 9. K. H. Ph.D. thesis of Texas, Austin, Clary, (1997) (Univ. TX). 29 Nielsen, E. S. & Kristensen,N. P. (1996)Invert. Taxon. 10, 1199-1302. 10. Brown, D. E., ed. Biotic Communities: Southwestern United States and (1994) Krenn,H. W. & Kristensen,N. P. (2000) Zool. Anz. 239, 179-196. Northwestern Mexico (Univ. of Utah Press, Salt Lake City). Pellmyr,0. (1999) Syst.Entoniol. 24, 243-271. 11. Simpson, G. G. (1953) The Major Features of Evolution (Columbia Univ. Press, 3 Cohen, S., Bronner,M., Kuttner,F., Jurgens,G. & Jackle,H. (1989) Natule New York). (London)338, 432-434. 12. Heard, S. B. & Hauser, D. L. (1995) Hist. Biol. 10, 151-173. G., Irvine,S. M., Lowe,C., Roehl,H., L. S., Sherbon,B., 13. Schluter, D. (2000) The Ecology of Adaptive Radiation (Oxford Univ. Press, Panganiban, Corley, J., J. J., Walker,M., et al. Proc.Natl. Acad. Oxford). Grenier, Fallon, F., Kimble, (1997) Sci. USA 5162-5166. 14. Riley, C. V. (1872) Nature (London) 6, 444. 94, A. & T. Dev. Biol. 673-689. 15. Riley, C. V. (1873) Missouri St. Entomol. Annu. Rep. 5, 150-160. 34. Abzhanov, Kaufman, (2000) 227, W. A. & T. C. 16. Pellmyr, O., Leebens-Mack, J. & Huth, C. J. (1996) Nature (London) 380, 35. Popadic,A., Panganiban,G., Rusch,D., Shear, Kaufman, (1998) 155-156. Dev. GenesEvol. 208, 142-150. 17. Frack, D. C. (1982) M.S. thesis (California State Polytechnic University, 36. Scholtz, G., Mittmann,B. & Gerberding,M. (1998) Int. J. Dev. Biol. 42, Pomona, CA). 801-810. 18. Pellmyr, 0. (1997) Ecology 78, 1655-1660. 37. Kopp,A., Duncan,I. & Carroll,S. B. (2000)Nature (London) 408, 553-559.

5502 I www.pnas.org/cgi/doi/10. 1073/pnas.072588699 Pellmyrand Krenn