Vol. 289 No. 5477 Pages 205-348 $8 Mi I' r ^ r^ -m 3 1 ?\\ ??^i. (iM I'! if '^??Ml vj ?;^ AMERICAN ASSOCIATION ?.L-OR. THE ADVANCEMINT OF S^Ci EVOLUTIONARY BIOLOGY Chewed Leaves Reveal Ancient Relationship God, the great British geneticist J. B. S. Hal- dane once remarked, must have "an inordi- nate fondness for beetles." And cejiain bee- tles have an tnordtnate and, it turns out, his- toric fondness for ginger plants. Paleontolo- gists have discovered how ancient this culi- nary preference really is by studying fossils of damaged leaves. The data help push back the time when a group of beetles called leaf beetles evolved their great diversity and demonstrate just how faithfiil some species can be to their favorite foods. The results are also convincing paleobotanists that they can sometimes glean more about their plant's ancient past from a chewed-up leaf fossil than from a pristine one. On page 291, paleobotanist Peter Wilf of the Uiiiversity of Michigan, Ann Arbor, Conrad Labandeira, a paleobiologist at the Smithsonian Institution's National Museum of Natural History in Washington, D.C., and their colleagues describe a new beetle fossil based not on traces of the insect skeleton? in fact, the insect itself never even shows up in the fossil record?but on the distinctive gouges the beetles left when they munched on 11 ginger leaves many millions of years ago. The chew marks of the newly described Cephaloleichnites strongi prove that leaf beetles underwent rapid evolution and diver- sification more than 65 million years ago? far earlier than the oldest fossils of insect bodies suggest?possibly taking advantage of (and perhaps influencing) the rapid diver- sification among flowering plants occurring at the same time. What's more, C. strongi represents the earliest known rolled-leaf beetle species, hundreds of which today still are picky eaters, preferring just one of the ginger- and heliconia-like plants in the Zingiberales order. For decades, ecology students have learned about this impressive array of beetle-plant pairings, in which different rolled-leaf species adopt the same lifestyle but on their own distinct host plant. This new work adds "a historical dimension to this emblem of tropical biology," says Brian D. Farrell, an insect evolutionist at Harvard University. Adds Phyllis Coley, a tropical ecologist at the University of Utah, Salt Lake City: "The beetles and the gingers are an extremely old and conservative pairing, which m turn suggests that each could have had profound selective effects on the other." As a young ecologist in the 1970s, Don- ald Strong?the fossil's namesake?could not help but notice the vast variety of rolled- leaf beetleis, whose larvae take up residence inside the young, curled leaves of gingers. NEWS OF THE WEEK heliconias, and their relatives, plants that thrive in the understories of tropical forests of the Western Hemisphere. In particular, he was enchanted by what the beetles did to the leaf itself Their damage becomes quite ap- parent as the leaf unfiirls and serves as a lasting reminder of a beetle long gone. "It was an issue of artistry, how beautifiil the damage was," recalls Strong, now at the Telltale jaws. From the characteristic chew marks left on fossilized (eaves, researchers have identified an ancient beetle and its fa- vorite food. Rolled-leaf beetles today still munch on ginger plants, as shown by the char- acteristic damage on this leaf from Panama. University of California, Davis. Over the next few decades. Strong docu- mented the specialized associations among different beetles and particular plant species. Eventually, he learned to identify a beetle species from the leaf's chew marks, which varied according to the size and shape of the particular beetle's jaws. Wilf came across Strong's research in 1998, when he and Labandeira were studying a different sort of insect damage?tiny fossil pellets, mysterious specks of fossilized mate- rial found on 53-million-year-old fossil leaves he had collected from Wyoming. Until that time, Wilf hadn't really noticed the chew marks. But when he and Labandeira took a second look at the leaves, "we realized the damage [seen by Strong in the modern leaves] matched beautifully with what we had," Labandeira recall?. Moreover, the fossil leaves looked very much like some modem gingers. Even after millions of years, says Wilf, "[the beetles] are eating the same thing, and they are doing it the same way." Soon Labandeira found even older leaves bearing the telltale signs of the rolled-leaf beetle. While working with Kirk Johnson at the Denver Museum of Natural History, La- bandeira noticed that some of Johnson's fos- sils, whose identity he didn't yet know, also had chew marks resembling C. strongi's. And they, too, turned out to be fossil gin- gers. Because these fossils came from a North Dakota deposit dating back to the Late Cretaceous, "we now know this insect is 20 million years older than if we just looked at body fossils," Wilf points out. These findings lend support to a theory proposed by Farrell in 1998. Farrell suggest- ed that most plant-eating beetles likely evolved in parallel to flowering plants and therefore were quite diverse during the di- nosaur's heyday (Science, 24 July 1998, p. ' 555). But until now, there has been little : supporting fossil evidence, as only one rele- vant beetle fossil exists from that time. Now researchers may be able to get around this lack of fossils by looking at insect damage instead, says Leo Hickey, a paleobotanist at Yale University; "The work shows the po- tential of an overlooked resource in [study- ing] the evolution of insects." Inspired by this new work, Hickey expects that he and his botanical colleagues will be giving their plant fossils a second look for signs of insect activity. Coley agrees, noting that "it seems that the use of fossil damage patterns to in- fer ecological and evolutionary relationships is quite powerful." -ELIZABETH PENNISI www.sciencemag.org SCIENCE VOL 289 14 JULY 2000 229 COVER A rolled-leaf hispine beetle (-5 mm long), Cephaloleia dor- salis, on a petiole of its host plant, Costus sp., in Chiriqui Province, Republic of Panama. Costus belongs to the order Zingiberales, con- sisting of gingers and related plants. Leaf damage inflicted on fossil gingers by ancient relatives of this beetle demonstrates that hispine beetles and their feeding associations with Zingiberales originated before the end of the Age of Dinosaurs. [Photo: D. M.Windsor] Timing the Radiations of Leaf Beetles: Hispines on Gingers from Latest Cretaceous to Recent Peter Wilf,^'^* Conrad C. Labandeira.^-^ W. John Kress,^ Charles L Staines,'* Donald M. Windsor,^ Ashley L AUen,^ Kirk R. Johnson^ Stereotyped feeding damage attributable solely to rolled-leaf hispine beetles Is documented on latest Cretaceous and early Eocene ginger leaves from North Dakota and Wyoming. Hispine beetles (6000 extant species) therefore evolved at least 20 million years earlier than suggested by insect body fossils, and their specialized associations with gingers and ginger relatives are ancient and phy- logenetically conservative. The latest Cretaceous presence of these relatively derived members of the hyperdiverse leaf-beetle clade (Chrysomelidae, more than 38,000 species) implies that many of the adaptive radiations that account for the present diversity of leaf beetles occurred during the Late Cretaceous, contemporaneously with the ongoing rapid evolution of their angiosperm hosts. Insects and flowering plants (angiosperms) tiires of terrestrial eeosystems (7). Diagnostie comprise most terrestrial biodiversity, and insect damage on fossil angiosperms is a their trophic associations are dominant fea- primary source of data for understanding the evolution of these associations and can also 'Museum of Paleontology and Department of Ceo- provide information complementary to insect logical Sciences, Universityof Michigan, Ann Arbor, Ml body fossils on the times of appearance of 48109-1079, USA. ^Department of Paleobiology, ^,^^?^^ lineages (2). Such insect damage is ^Department of Botany, ?'Department of Entomology, , , iiz- j' /T^N National Museum of Natural History, Smithsonian '^"?^^" ^'"i"?! exclusively from dicots (i, 4), Institution, Washington, DC 20560, USA. ^Depart- althougll monocots comprise ~22% of living ment of Entomology, University of Maryland, College angiosperm species (5) and are hosts to di- Park, MD 20742-4454, USA. ??Smithsonian Tropical verse groups of herbivorous insects (6, 7). Research Institute, Apartado 2072, Balboa-Ancon, Re- . .i, u . . j- j ? .-? u .. ,,. ,. Z, ,. .. ? r ..u j <- Amone the best studied associations between public of Panama. 'Department of Earth and Space " Sciences, Denver Museum of Natural History, Denver, insects and monocots is the specialized feed- CO 80205, USA. ing of rolled-leaf hispine beetles (family *To whom correspondence should be addressed. E- Chrysomelidae, subfamily Hispinae, tribes mail: pwilf@umich.edu Ccphalolciini and Arcscini) in the scmi- www.sciencemag.org SCIENCE VOL 289 14 JULY 2000 291 REPORTS aquatic shaded habitat provided by the rolled juvenile leaves of gingers, heliconias, and their relatives (order Zingiberales) in under- stories of Neotropical forests (Fig. 1) (8-13). The feeding marks of larval rolled-leaf hispines are stereotyped (9) (Fig. 1) and re- main intact on the mature unrolled leaves, increasing their potential for fossilization. The family Chrysomelidae, or "leaf bee- tles," has ~38,000 described species (14) and a possible total diversity of > 60,000 species (15). Most extant leaf beetles con- sume angiosperms, indicating a series of adaptive beetle radiations (7). The subfamily Hispinae (-6000 species) (7, 13, 15-17) is considered to be among the more derived and specialized groups within the Chrysomelidae (6, 7). The Hispinae and its putative sister group (~5000 species) (Fig. 2) comprise a clade that includes most extant species of monocot-feeding beetles (18). The body-fossil record of leaf beetles is virtually nonexistent during the Late Creta- ceous (7, 19), the time interval known for rapid evolution and diversification of angio- sperms (20), and the record of most angio- sperm-feeding Chrysomelidae is confined to the Cenozoic (7). The first appearance of Hispinae, in particular, is in the middle Eo- cene, and the rolled-leaf hispines have no fossil record (Fig. 2). This lack of temporal resolution limits understanding of the timing of chrysomelid radiations in relation to the evolution of angiosperm host plants, whose Cretaceous fossil records are far more com- plete than those of leaf beetles (5, 21, 22). Here, we report diagnostic feeding pat- terns, of the type documented for larvae of living rolled-leaf hispines in Central America (9), on 11 specimens of latest Cretaceous and early Eocene Zingiberopsis (Fig. 1). This well-described leaf genus, a fossil member of the ginger family (Zingiberaceae), is known from Late Cretaceous through earliest Oligo- c?ne strata of North America and from the early Late Cretaceous of Germany (23-26). The nearest living relative o? Zingiberopsis is considered to be the Asian genus Alpinia (24) (Fig. 3). Of the 11 insect-damaged specimens studied, 7 were Z. isonervosa from the early Eocene Wasatch Formation, Great Divide Basin, southwestern Wyoming (26-28). The remainder were three specimens of Z. attenu- ata, from the latest Cretaceous Hell Creek Formation, and a single specimen of Z. ison- ervosa from the early Eocene Camels Butte Member of the Golden Valley Formation; all four specimens are from the Williston Basin, southwestern North Dakota (28). The damage consists of individual (Fig. IE) or sequential (Fig. 1, C and F through I) linear feeding strips that are bounded by reaction tissue and have asymmetrically rounded termini, as de- scribed in detail below (29). We propose the ichnotaxon Cephaloleichnites strongi, gen. et Fig. 1. Recent and fossil [Cephaloleichnites strongi) hispine damage on Zingiberales (29). (A) is live; (B) and (D) are pressed specimens from the U.S. National Herbarium; (C), (E), (H), and (I) are from the early Eocene; and (F) and (C) are from the latest Cretaceous [28, 29). (A) Chelobasis perplexa Baly larva feeding on a leaf of Heliconia curtispatha Petersen (collected in Chiriqui Province, Panama). The arrows indicate damage trails with irregular margins that are deployed perpendicular to leaf venation. (B) Hispine damage of the type noted by the arrows in (A) on Heliconia vaginalis Bentham [U.S. National Herbarium (US) 3134380, collected in Costa Rica]. (C) C. strongi (holotype) on Zingiberopsis isonervosa Hickey (USNI^ 498174). (D) Hispine damage on Renealmia cernua (Swartz) l^acbride (Zingiberaceae), a close relative of Zingiberopsis (Fig. 3) (US 1153643, collected in Panama). Extended linear slot feeding is visible. (E) C. strongi, single slot of the type shown in (D) (USNM 498168). (F and C) C. strongi on Z. attenuata Hickey and Peterson [DMNH 19957, (F); DI^NH 19959, (G)]. (H and I) C. strongi on Z. isonervosa [USNI^ 509718, (H); USNM 498169, (I)]. Scale bars in all panels equal 5 mm, except in (G), where the scale bar is 1 mm. sp. nov., for the fossil insect damage (29). The best fit of current phylogenetic data to the fossils suggests a basal member of a derived group, the Hispinae, feeding on a derived monocot host (Figs. 2 and 3). An adaptive trajectory within the phylogeny of Hispinae and their close relatives is depict- ed in Fig. 2, which starts on aquatic and semiaquatic dicots and then shifts to mono- cot host plants (stages 1 to 4 in Fig. 2) (18). C. strongi documents the extension of the semiaquatic life-style inland to the wet rolled-leaf habitat of Zingiberales (stage 5 in Fig. 2). Host shifts by higher hispine taxa occurred on terrestrial monocots and, for the "cassidoid" group, on dicots (stages 6 to 8 in Fig. 2). The present-day occurrence of rolled-leaf hispines on six of the eight families of Zin- giberales (13) raises the question of the order of colonization within Zingiberales. Plant chemistry is thought to be a primary con- straint and selective force on the host shifts of Chrysomelidae and other herbivorous beetles (30). The Zingiberaceae as a group possess well-developed ph5'tochemistry, and several compounds with potentially defensive uses have been extracted from Alpinia leaves in particular, including tannins, phenols, alka- loids, and diverse terpenes (31). In contrast. leaves of Heliconia, a relatively basal mem- ber of the Zingiberales (Fig. 3) that is host to a high diversity of rolled-leaf hispines (8), are notably lacking in defensive compounds, and experimental data show negligible effects of Heliconia chemistry on the larval develop- ment of rolled-leaf hispines (11). Corre- spondingly, we have also observed, in her- barium collections, a relatively low frequen- cy and intensity of hispine damage on Zin- giberaceae in comparison to Heliconia. The preceding evidence suggests an initial colo- nization of basal, chemically "simple" Zin- giberales, which led to the diverse associa- tions with living Heliconia, followed by adaptive radiations of specialized hispines on the Zingiberaceae by the Maastrichtian or earlier. C. strongi predates the body-fossil record of Hispinae by ~20 million years, document- ing the Cretaceous origins of the group (Fig. 2). As the fossil records of many living lin- eages of monocots begin in the Campanian and Maastrichtian (5), our data demonstrate the presence and trophic activity of derived, specialized, monocot-feeding beetles near the time of the first appearances of present-day host groups. In addition, the recent discovery of a fossil sagrine beetle (19) indicates the presence of the sister group to the hispines by 292 14 JULY 2000 VOL 289 SCIENCE www.sciencemag.org REPORTS BEETLES 6000 spp. I D Insect damage (Dthis report) Fig. 2. Hypothesized evolutionary coloniza- tion of angiosperms by hispine beetles and their immediate ancestors, with the corresponding fossil record of bee- tles and their feeding damage (79, 36). At the left is a phyloge- ny of hispine tribes (blue) (73) and subfam- ilies and tribes of its putative sister group (green) (37), with an empty branch repre- senting all other Chry- somelidae; dashed lines indicate groups without published phylogenies, inserted on the basis of morphological evidence (38). Approximate num- bers of described spe- cies (spp.) are indicated for these three branch- es, using (74) for the total of 38,000 and (7) for the blue and green clades. The two tribes of rolled-leaf hispines are in capital letters. Rele- vant body fossils of in- sects are almost entirely confined to Cenozoic Lagerst?tten. At the top right is a cladogram of all major monocot lineages and several representative clades of basal dicots, which is a compromise topology among recently hypothesized evolutionary relationships based on both molecular and morphological charac- ters (39-41). The lower right indicates dominant (red squares) and subdominant (orange squares) plant hosts for chrysomelid clades at the left (6, 42-44); numbered red clusters represent inferred major colonization stages. The matrix reflects larval herbivory, almost all of which is deployed as external feeding or leaf mining. The overall trajectory of primitive aquatic dicot to advanced PLANTS ininmmwwwinmwwinmwwininmviin QuOninniniraninniraninnirarasQra # Insect body-fossils iOOO spp. ^ ?^ ? ? ?^ "cassidoid" Hispinae ^ Hemisphaerotini Prosopodontini Alurnini Cryptonychini Hispini Delocraniini Oediopalpini Uroplatini # Chalepini Sceloenoplini --ARESCINI ? CEPHALOLEIINI ?^ Bruchinae ^ Ambiycerinae ^ Pachymerinae ~ Sagrinae ^ Criocerinae -^ Haemoniini -^^9 Donaciini 27000 spp. I^9 Plateumarini ? Other Chrysomelidae Cam. |Maa.| 80 70 Pal. ?r- 60 Eocene ?\ r 50 4t Olig. Miocene I? 20 ?r- 10 Epoch O Ma ASSOCIATIONS monocot to core eudicot host colonization is indicated by the stippled arrow; secondary colonizations of core eudicots (43), as supported by beetle phylog- enies, are designated by smaller arrows (6,45). The colonization of core eudicots by "cassidoid" hispines is primary (43, 44). The actual history of colonization is undoubtedly more complex than depicted, and the time scale refers only to fossil occurrences, not to branching events. The blank sections of the time scale are "Pliocene" and "Pleistocene," from left to right. Cam. = Campanian; l^aa. = l^aastrichtian; Pal. = Paleocene; Olig. = Oligoc?ne; Donac = Donaciinae; C = Criocerinae; S = Sagrinae; Bruch = bruchoid complex. the Campanian (Fig. 2). Taken together, the Cretaceous hispine and sagrine occurrences indicate a high likelihood that many other clades of leaf beetles evolved well before the terminal Cretaceous. Angiosperm diversity exceeded that of other groups of land plants by the early Late Creta- ceous (20). The rapid evolution of angiosperms continued throughout the Late Cretaceous (22), and 44% of extant angiosperm orders have Cretaceous fossil records, including most living lineages (21). Thus, Cretaceous radiations of leaf beetles occurred during an extended inter- val of evolutionary innovation for angiosperms, suggesting the possibilities of plant-beetle co- evolution or of adaptive beetle radiations that closely followed the diversification of angio- sperms. Supporting the latter hypothesis is Far- rell's contrast of the diversities of sister groups of gymnosperm- and angiosperm-feeding bee- tles (7), leading to his estimate that radiations of beetles on angiosperms were responsible for the evolution of ~ 100,000 living beetle species. Rolled-leaf Hispinae and Zingiberales have maintained a stereotyped, highly spe- cialized plant-animal interaction in the New World for >66 million years, surviving the mass extinctions of plants at the end of the E m o B O S ??.-o o ?? := ? -I CO ? (O o u Sg?.?g.S|20?, and ~18?C for the relevant portions of the Hell Creek [K. R. Johnson and P. Wilf, CeoL Soc. Am. Abstr. Progr. 29, 432 (1996)], Wasatch [26), and Golden Valley [23] formations, respectively. 29. Genus: Cephaloleichnites, gen. nov., subfamily Hispinae. Genotypical species: Cephaloleichnites strongi, sp. nov. Generic diagnosis: The genus pertains to fossil traces of insect feeding consisting of linear strips, each con- fined entirely within the space between adjacent par- allel veins such that leaf-tissue strata between parallel veins are removed and only the upper epidermis typi- cally remains (Fig. 1). Strips are bordered by dark reac- tion tissue of the host plant. Terminations of strips are usually asymmetrically rounded. The average strip length is 2.1 mm (a = 0.83 mm, minimum = 0.81 mm, maximum = 6.3 mm, n = 279). Eocene strip lengths are shorter (mean of 1.9 mm for the Wasatch Forma- tion, n = 209; mean of 2.0 mm for the Golden Valley Formation, n = 42} than those from the Cretaceous (mean of 3.5 mm, n = 28). Strips are occasionally single (Fig. IE), nearly always consecutive, and characterized by series of strips occupying adjacent pairs of parallel veins so as to form a continuous and en ?chelon dam- age field. Single strips and consecutive strips can co- occur on a single specimen (USNM 498168), as they do on modern examples (Fig. ID). The series of end points of consecutive strips is very roughly linear, resulting in an overall squarish or otherwise quadrilateral feeding feature that has a ragged irregular margin. The angle of the feature's margin to the parallel veins of the host plant is typically perpendicular but can be angled up to 30? from perpendicular. The maximum number of con- secutive strips found is 34, on the holotype. Species diagnosis: Diagnosis is the same as that for the genus, because of monotypy. Repository: All type and re- ferred material is housed in the paleobotanical type collections of the USNM (National Museum of Natural History, Smithsonian Institution) and the DMNH. Ho- lotype: USNM 498174 (Fig. 1C). Type locality: USNM loc. 41352. Referred material: DMNH 19957, 19959, and 19960 (DMNH loc. 2092); USNM 498168 (USNM loc. 41362), 498169 through 498173 (USNM loc. 41352), and 509718 (USNM loc. 14048). Etymology: Cephaloleia Chevrolat is the only extant genus of rolled-leaf Hispinae known to feed on Zingiberaceae today (9), although both rolled-leaf tribes, the Cephalo- leiini and the Arescini, generate similar leaf damage on other Zingiberales [?chnos: trail, track (Greek); strongi: named for D. R. Strong Jr., for his seminal papers on the modern analog association]. Discussion: The fossil and modern damage are equivalent, and only the rolled-leaf hispines are known to produce the relevant damage patterns on living Zingiberales. Cephaloleichnites indi- cates a probable tribal affinity but not a formal tribal classification. C. strongi in all probability spatiotempo- rally represents more than one larval beetle species. Feeding is accomplished by "scraping the ventrally- directed, scoop-shaped, toothed mouthparts reciprocal- ly across the plant surface" (9, p. 158). Adult hispines leave similar damage on the same hosts as larvae, but the margin of the damage field typically is more smooth (9). The fossil damage was first noted in table 1 of (27, p. 2154} as "strip-feeding between secondary veins [Zingiberopsis)." This ichnotaxonomic description is provided for by W. D. L. Ride and others [W. D. L. Ride ei a/., Eds., International Code of Zoological Nomencla- ture (International Trust for Zoological Nomenclature, London, ed. 4, 1999), article 1.2.1]. 30. J. X. Becerra, Science 276, 253 (1997); D. J. Futuyma and S. S. McCafferty, Evolution 44, 1885 (1990); B. Meurer-Grimes and G. Tavakilian, Bot. Rev. 63, 356 (1997); A. Kopf ei a/.. Evolution 52, 517 (1998). 31. V. L. M. Mendon?a, C. L. A. Oliveira, A. A. Craveiro, V. S. Rao, M. C. Fonteles, Mem. Inst. Osv/aldo Cruz 86, 93 (1991); K. S. Ngo and G. D. Brown, Phytochemistry 47, 1117 (1998). 32. K. R. Johnson and L. J. Hickey, in Global Catastrophes in Earth History: An Interdisciplinary Conference on Impacts, Volcanism, and Mass Mortality, V. L. Sharp- ton and P. D. Ward, Eds. (Geological Society of Amer- ica, Boulder, CO, 1990), pp. 433-444. 33. C. H. Lear, H. Elderfield, P. A. Wilson, Science 287, 269 (2000). 34. D. J. Futuyma and C. Mitter, Philos. Trans. R. Soc. London Ser. B 351, 1361 (1996); T. H. Hsiao and J. M. Paste?is, in Advances in Chrysomelidae Biology, M. L. Cox, Ed. (Backhuys, Leiden, Netherlands, 1999), vol. 1, pp. 321-342. 35. A. T. Peterson, J. Sober?n, V. S?nchez-Cordero, Sci- ence 285, 1265 (1999). 36. M. E. Collinson and H. J. Gregor, Tertiary Res. 9, 67 (1988); J. A. Santiago-Blay, in Novel Aspects of the Biology of Chrysomelidae, P. H. Jolivet, M. L. Cox, E. Petitpierre, Eds. (Kluwer, Dordrecht, Netherlands, 1994), pp. 1-68; G. O. Poinar Jr. and R. Poinar, The Amber Forest (Princeton Univ. Press, Princeton, NJ, 1999); S. B. Archibald and R. W. Mathewes, Can. J. ZooL, in press); and other sources. Fossil hispine damage reported here (yellow squares in Fig. 2} is placed on the Cephaloleiini branch for convenience, although the Arescini cannot be excluded as a remote possibility for these ancient feeders. 37. The phylogeny is based on (7) for subfamilial rela- tionships and on (77) and work by I. S. Askevold [Can. J. ZooL 68, 2135 (1990)] for hispine and donaciine tribal groupings, respectively. Resolution of subfam- ilial relationships is based on morphological and mo- lecular data, donaciine relationships are based on morphological data only, and hispine relationships are based on molecular data only. 38. M. Schmitt, ZooL Beitr. 29, 35 (1985); K. Suzuki, in Chrysomelidae Biology, P. H. Jolivet and M. L. Cox, Eds. (SPB Academic, Amsterdam, 1996), voL 1, pp. 3-54; (6, 75, 76, 42). 39. W. J. Kress, L. M. Prince, W. J. Hahn, E. A. Zimmer, unpublished data. 40. M. W. Chase et aL, in Monocotyledons: Systematics and Evolution, P. Rudall et aL, Eds. (Royal Botanic Gardens, Kew, UK, 1995), pp. 109-138; J. I. Davis, Syst. Bot. 20, 503 (1995); D. Stevenson and H. Loconte, in Monocoty- ledons: Systematics and Evolution, P. Rudall ei a/., Eds. (Royal Botanic Gardens, Kew, UK, 1995), pp. 543-578; T. J. Givnish, T. M. Evans, J. C. Pires, K. J. Sytsma, MoL Phylogenet. EvoL 12, 360 (1999). 41. Although there is agreement concerning the compo- sition of the major lineages of monocots, some un- certainty still exists as to the exact topology of the Liliales, Asparagales, Dioscoreales, Pandanales, Tri- uridales, and Petrosaviales. "Poales and allies" in- cludes the cattails, pineapples, sedges, rushes, grass- es, and relatives. The clade from the Proteales con- sumed by Donaciini and Plateumarini is the Nelum- bonaceae, which are aquatic. 42. L. Borowiec, in Biology, Phylogeny, and Classification of the Cole?ptera: Papers Celebrating the 80th Birth- day of Roy A. Crowson, J. Pakaluk and S. A. Slipinski, Eds. (Polish Academy of Sciences, Warsaw, 1995), pp. 541-558; D. M. Windsor, unpublished data; C. L. Staines, unpublished data. 43. L. G. E. Kalshoven, Tijdschr. EntomoL 100, 5 (1957). 44. P. H. Jolivet, BulL Mens. Soc. Linn. Lyon 58, 297 (1989). 45. L. Borowiec, PoL Pismo EntomoL 57, 3 (1987). 46. W. J. Kress, in Monocotyledons: Systematics and Evo- lution, P. J. Rudall ei a/., Eds. (Royal Botanic Gardens, Kew, UK, 1995), pp. 443-460; W. J. Kress, L. M. Prince, K. J. Williams, unpublished data. 47. We thank W. Crepet, B. Farrell, and two anonymous colleagues for reviews; T. Baumiller, R. Burnham, D. Fisher, D. Furth, L. Hickey, R. Horwitt, and S. Wing for reviews of drafts; M. Guerra for photography (Fig. 1 A); I. L?pez for assistance with herbarium material; F. Marsh for rendering Fig. 2; and B. Miljour for assistance with Fig. 1. P.W. was supported by a Smithsonian Institution Postdoctoral Fellowship, the Smithsonian's Evolution of Terrestrial Ecosystems Program (ETE), and the Michigan Society of Fellows. C.C.L. received support from the Smithsonian Walcott Fund, and W.J.K. received support from a Smithsonian Scholarly Studies Grant. This is ETE contribution number 74. 29 March 2000; accepted 18 May 2000 294 14 JULY 2000 VOL 289 SCIENCE www.sciencemag.org