Pacific Science (1997), vol. 51, no. 4: 440-449 © 1997 by University of Hawai'i Press. All rights reserved

Herbivorous Insects and the Hawaiian Silversword Alliance: Coevolution or Cospeciation?l

GEORGE K. RODERICK2

ABSTRACT: Numerous groups of herbivorous insects in the Hawaiian archipelago have undergone adaptive radiations. R. C. L. Perkins collected and documented species in nearly all of these groups. In this study I tested whether patterns of host plant use by herbivorous insects can be explained by host plant history. I examined a group ofinsects in the planthopper genus Nesosydne (Hemiptera: Delphacidae) that feed on plants in the Hawaiian silversword alliance, many of which are endangered or threatened. For these Nesosydne species feeding on the silversword alliance, mitochondrial DNA sequence data revealed a statistically significant pattern of cospeciation between these insects and their hosts. These planthoppers are highly host-specific, with each species feeding on only one, or a few closely related, plant species. Patterns ofhost plant use across the plant lineage, as well as within extensive hybrid zones between members of the silversword alliance, suggest that planthopper diversification parallels host plant diversification. Data collected thus far are consis­ tent with, but do not directly demonstrate, reciprocal adaptation. For other herbivo­ rous insects associated with members of the Hawaiian silversword alliance, patterns of host plant use and evolutionary history are not yet well understood. However, cospeciation appears not to be universal. For example, endemic flies in the family Tephritidae (Diptera) are less host-specific and demonstrate host-switching. Research is under way to reveal the mechanisms associated with cospeciation and host­ switching for different insect groups associated with the Hawaiian silversword alliance.

INSECT LINEAGES AT the generic and family lev­ Cospeciation is the matching of speciation els appear to be largely conservative with respect events in two lineages, such that the two phylog­ to their host affiliations, and many species in enies resemble one another (see Brooks 1979, such lineages are highly host-specific (Dethier 1988, Mitter and Brooks 1983, Hafner et al. 1954, Ehrlich and Raven 1964, Farrell and Mit­ 1994, Page 1995a). Cospeciation of plants and ter 1993, but see Dobler et al. 1996). Current their herbivorous insects may increase herbivore patterns of host plant use by herbivorous insects diversity as insects track and diversify with their can be explained by one of two hypotheses, both hosts. Coevolution can lead to cospeciation but of which can lead to greater herbivore diversity: is not required. For example, major vicariant cospeciation with host plants and host-plant events to which both lineages respond, followed switching. by allopatric speciation, could also produce a pattern of cospeciation. Or one group may I This research was supported by the University respond to vicariance, with the other lineage Research Council of the University of Hawai'i and the U.S. "following" again without the need for strict Fish and Wildlife Service. Additional support for our labora­ tory is provided by grants from NSF, USDA, California coevolution. Department of Food and Agriculture, International Rice By contrast, host-plant switching is a change Research Institute, Rockefeller Foundation, and the Pacific ofhosts (see Futuyma 1983a,b, Thompson 1994) Biomedical Research Center, University of Hawai'i. Manu­ other than would be predicted by the host phy­ script accepted 3 February 1997. logeny, such that the two phylogenies are no 2 Center for Conservation Research and Training, 3050 Maile Way, Gilmore 409, University of Hawai'i at Manoa, longer congruent. Host-switching can result in Honolulu, Hawai'i 96822. greater herbivore diversity if, following a switch 440 Herbivorous Insects on Hawaiian Silverswords-RoDERlCK 441 to a new host, the herbivore becomes sufficiently concurrently. However, for many insect/plant isolated and divergent from the species on the relationships studied, especially in North original host. However, factors responsible for America, the radiations of plants and insects generating the diversity of herbivorous insects have not been coincident, usually with the plant relative to their hosts are less clear. Compilations group becoming established following retreat of ofresearch on herbivorous insects and their hosts the glaciers and the insects spreading northward indicate that although a few insect radiations do at some later time (Farrell and Mitter 1993, Funk appear to be tightly correlated with radiations et al. 1995). For such interactions in mainland of their hosts, most insect radiations show evi­ tropical regions, which are older than temperate dence of host-switching (Mitter et al. 1988, ones, historical patterns of host association are 1991, Farrell and Mitter 1993, Funk et al. 1995). likely also to be obscured by extinctions and Recently, it has been suggested that hybrids incomplete knowledge (B. Farrell, pers. comm.). between host species may be associated with Such circumstances may bias evidence away herbivorous insect diversity and, by extension, from cospeciation, likely discounting its impor­ parasite diversity (Floate and Whitham 1993, tance in the early stages of diversification of a Strauss 1994, Whitham et al. 1994, Christensen plant lineage. To know about the relative ages et al. 1995). Much ecological and genetic data of the cospeciation events requires some knowl­ have demonstrated that hybrid zones can repre­ edge of the historical frameworks of both the sent very different environments for herbivorous insect and plant radiations. insects (see Boecklen and Spellenberg 1990, Many radiations of herbivorous insects are Aguilar and Boecklen 1992, Floate et al. 1993, endemic to Hawai'i (see Simon 1987, Howarth Floate and Whitham 1994, Fritz et al. 1994, and Mull 1992, Nishida 1994, Asquith 1995, Roderick and Metz 1997). Hybrid hosts not only Eldredge and Miller 1995, Miller and Eldredge create new adaptive zones, but also may allow 1996, Roderick and Gillespie in press) and repre­ specialized parasites an escape from the evolu­ sent a unique opportunity to unravel the popula­ tionary dead end-hybrids may enable parasites tion genetic and phylogenetic processes that to increase their host range, possibly leading have led to current patterns of host plant affilia­ to genetic diversification and speciation. This tion and are responsible for the diversity of asso­ scenario has been termed the "hybrid bridge" ciated insects. The extreme isolation and hypothesis (Floate and Whitham 1993). Hybrid historical framework provided by the Hawaiian hosts thus can provide a means for host­ archipelago makes possible the set of features switching. necessary for the test of hypotheses to account Distinguishing between the relative impor­ for patterns of herbivore host affiliations and tance ofcospeciation and host-switching is diffi­ associated diversity. In particular, island and vol­ cult for several reasons. By necessity, much of cano age can be used to identify the age of our information concerning patterns ofcospecia­ particular insect/plant associations, indepen­ tion or coevolution must come from inferring dently from any information gained through past events from current observations. Recent molecular data. In this study, I examined the developments in both molecular biology and the­ evidence for cospeciation and host-switching for ory have made it possible to compare recon­ species of planthoppers in the genus Nesosydne structed phylogenies of pairs of interacting (Hemiptera: Delphacidae) that feed on the lineages and to test for significance of cospecia­ Hawaiian silversword alliance. Recently, Bald­ tion (Mitter and Brooks 1983, Brooks 1988, Mit­ win and Robichaux (1995) generated a phyloge­ ter et al. 1991, Maddison and Maddison 1992, netic hypothesis for the history of the species Moran and Baumann 1994, Funk et al. 1995, within the silversword alliance that makes this Page 1995a, b). However, reconstructing past work possible. associations between species can be more prob­ The current threats to biological diversity lematic (Mitter et al. 1988, 1991, Wiegmann et have necessitated the understanding ofthe forces al. 1993). For example, if cospeciation were to responsible for both its generation and demise occur, it demands that both insect and plant (or (Wilson 1988, 1996). Nowhere on earth is the parasite and host) have the potential to radiate extinction crises more acute than in the Hawaiian 442 PACIFIC SCIENCE, Volume 51, October 1997

archipelago (Gillespie et al. 1997, Liebherr and of planthoppers and their host plants; (2) it gives Polhemus 1997, Gillespie 1997). As such, insight into the degree of host specificity; and Hawai'i becomes a model system for the study (3) it establishes the relatedness between paren­ of both the generation of biodiversity and its tal plant species associated with each plant conservation. hybrid zone. Nesosydne PLANTHOPPERS. The genus Neso­ sydne in Hawai'i contains at least 80 species (Zimmerman 1948). In contrast to other delpha­ MATERIALS AND METHODS cid planthoppers that are mainly grass feeders Study Organisms (Denno and Roderick 1990), species within Nes­ osydne in Hawai'i feed on plants in an aston­ SILVERSWORD ALLIANCE. The silversword ishing 28 plant families (Figure 1) (Zimmerman alliance in Hawai'i comprises 28 species, pre­ 1948, Wilson et al. 1994). This diversity ofhost sumably with one common ancestor (Baldwin et plants may be explained in part by the fact that al. 1991, Baldwin and Robichaux 1995, Baldwin Nesosydne is likely polyphyletic and may repre­ 1997). This radiation is among the most well sent several independent colonizations of studied of all plant lineages in Hawai'i (see Hawai'i with subsequent radiations within the Wagner and Funk 1995), with published works archipelago (Asche 1997). Despite this diversity on ecology, physiology, systematics, conserva­ of host plants, nearly all Nesosydne species are tion status, and hybridization (Carr 1987). Carr highly host-specific, with 88% of species using and colleagues (Carr 1985, 1990a,b) have inves­ plants within a single family and 77% using tigated the extent of hybridization between a single plant species (Figure 2). At least 15 members of the silversword alliance and have documented that many, if not most, members of Number of Nesosydne Species the silversword alliance form natural hybrids in o 5 10 15 20 the field. Hybrid zones differ in (1) the plant Asteraceae Gesneriaceae species involved; (2) the relatedness of plant Campanulaceae species that hybridize; (3) the range ofecological Urticaceae conditions occupied; and (4) the extent of over­ Fabaceae Rubiaceae lap between the hybrids and one or both parental Lamiaceae species. Hybrids and potential Fls have been Epacridaceae identified by leaf size and shape (Carr 1985), Liliaceae Amaranthaceae and recently by genetic data (Friar et al. 1996, Blechnaceae V. Caraway and C. Morden, unpubl. data). Gunneraceae Recently, Baldwin and colleagues (Baldwin Hydrangeaceae Pandanaceae et al. 1991, Baldwin and Robichaux 1995, Bald­ Arecaceae win 1997) used molecular genetic data to gener­ Brassicaceae Convolvulaceae ate a hypothesis of evolutionary relationships Davalfiaceae among members of the alliance. The alliance Dicksoniaceae appears to be monophyletic and to contain dis­ Geraniaceae Lythraceae tinct clades within Hawai'i. Divergence among Oleandraceae extant species is likely in the range of 4-6 myr, Piperaceae or no older than the age of Kaua'i (Baldwin and Poaceae Polygonaceae Robichaux 1995, Baldwin 1997). Conflicting Smilaceae evidence from nuclear, karyotype, and cyto­ Solanaceae plasmic DNA data suggests that some species Stilaginaceae may be of hybrid origin. The existence of a phylogenetic hypothesis for plant species in the FIGURE 1. Number of Hawaiian Nesosydne species recorded on species in each plant family (compiled from silversword alliance is an essential element in Zimmerman 1948, Wilson et al. 1994). Number of species the analysis presented here for several reasons: adds to more than total for Hawai'i because some planthop­ (1) it provides the basis for tests of cospeciation pefs feed on plant species in more than one family. Herbivorous Insects on Hawaiian Silverswords-RoDERICK 443

A. B. 80 80 70 70 (J) Q) 60 '0 60 Q) c. 50 50 C/) '0 40 40 ~ Q) .c 30 30 E ::::l 20 20 Z 10 10

0U--..J.J.L--UL--.....L....---L_--l==.l.-ff-...l==J OLL.-....l.LJ.---LU-.....I.L.c='----Lc=~--U::= 1 2 3 4 5 6 15 1 2 3 4 5 6 18 Number of Plant Families Number of Plant Species

FIGURE 2. Host specificity of single Nesosydne species in Hawai'i. (A) Distribution of number of plant families on which single species have been recorded. Total plant families recorded in Hawai'i = 28. (B) Distribution of number of plant species on which single Nesosydne species have been recorded. Nesosydne species for which plant families (A) are recorded. n = 71; and for which plant species (B) are recorded. n = 66.

Nesosydne species are reported to be found only divergence, Kimura (1980) 2-parameter, and on plant species within the Hawaiian silversword Tamura-Nei (1993). A phylogeny was recon­ alliance (Zimmerman 1948, Swezey 1954). Cur­ structed using parsimony (PAUP [Swofford rently, it is not known whether these 15 species 1993]). Bootstrapping was used to provide a represent their own monophyletic clade. level of confidence associated with each node. Outgroups included another Hawaiian delphacid planthopper, Nesosydne koae, which feeds on Host Plant Associations koa (see O'Connell 1991), and a delphacid from For the insects examined in this study, host the U.S. mainland, Prokelisia marginata (Roder­ plant associations for planthoppers on the sil­ ick 1987, Denno et al. 1997). versword alliance were established by field col­ lections of developing nymphs and the presence of male and female adults. Planthoppers were Cospeciation between Planthoppers and collected using an aspirator (puter) and sweep­ Host Plants ing. Insects were frozen at -80°C shortly A test of cospeciation was conducted using after collection. TreeMap (Page 1995b). This method and the corresponding computer program makes explicit Planthopper History the relationship between the host and "parasite" (here, insect herbivore) trees and allows a visu­ A 441 base-pair region ofmitochondrial cyto­ ally intuitive representation of that history (Page chrome oxidase I DNA was amplified using 1995a). In brief, reconstructions that maximize primers CI-J-1751 'Ron' and CI-N-2191 the number of cospeciation events are consid­ 'Nancy' (designed by R. Harrison laboratory ered to have the greatest explanatory power and [Simon et al. 1994D. Here, I present data from are preferred over reconstructions with fewer 20 individuals in six Nesosydne species. The cospeciation events. In the method, host frequency of transitions and transversions was switches are also incorporated as an explanation determined for the group using several genetic of the observed pattern ofhost-"parasite" associ­ distances: uncorrected pairwise percentage ations. The significance of the observed fit 444 PACIFIC SCIENCE, Volume 51, October 1997 between host and parasite trees can be evaluated those feeding on the silversword radiation, the by comparison with the distribution of the same group of species examined here was supported measure of fit for random trees (Page 1995a). by multiple synapomorphies that distinguished The program does place several constraints on it from N. koae (Figure 3). The time frame sug­ the data. For example, each "parasite" may only gested here for the diversification of Nesosydne be associated with one host, and the trees must species that feed on the silversword alliance is be completely resolved. consistent with a single origin of these insects The planthopper phylogeny was mapped on in Hawai'i and corresponds to the age of the to Baldwin and Robichaux's (1995) phylogeny silversword alliance. for the silversword alliance based on sequences ofnuclear ribosomal DNA. A randomization test (by "randomizing" the planthopper tree using Cospeciation between Planthoppers and TreeMap [Page 1995b]) was used to test the Host Plants significance of the observed level of cospecia­ tion between planthoppers and their plant hosts. The comparison ofplanthopper and host plant phylogenies resulted in five cospeciation events (noted by letters in Figure 3). The randomization RESULTS test shows that this number of cospeciation events is significant (P < .01). Note that the Planthopper Host Plant Associations method identifies "cospeciation" events for the individual planthoppers collected on Dubautia Planthoppers have now been collected on 13 raillardioides and D. paleata, and D. ciliolata out of 28 members of the silversword alliance. and D. scabra, even though the planthoppers Where a planthopper species has been recorded collected on each species pair are identified as on more than one host species, the hosts were the same species (see Figure 3). The comparison closely related and/or hybridize. These host of phylogenies provides no evidence of recent records suggest that planthopper species feeding host-switching. Whether the pattern of cospecia­ on the silversword alliance are highly host spe­ tion will hold when all species ofNesosydne are cific to either single plant species or closely included is one objective of current research. related species. Planthoppers were also collected from five extensive hybrid zones (Figure 4). Baldwin and Planthopper History Robichaux's (1995) data provide information on the relatedness of the plant species involved in The 441 base-pair piece of cytochrome oxi­ each of these hybrid zones. In two hybrid zones dase I amplified was one codon insertion longer between closely related plant species a single than that ofDrosophila yakuba (Clary and Wols­ planthopper species spanned each entire hybrid tenholme 1985). Transitions were approximately zone: D. paleata and D. raillardioides (Kaua'i) double transversions for the range of genetic and D. ciliolata and D. scabra (Hawai'i). In two distances encompassing the Nesosydne plant­ other hybrid zones between close plant relatives, hoppers feeding on the silversword alliance, and a single planthopper species was associated with both transitions and transversions increased lin­ only one parental plant species; in one of these, early over this range of genetic distances. The the planthopper also occurred on all identifiable linear increase indicates that cytochrome oxi­ hybrids, and in the other the planthopper's range dase I is a good candidate for the evolutionary was restricted to the parental species: D. ciliolata relationships investigated here and that both and D. arborea (Hawai'i) and D. menziesii and transitions and transversions contain useful D. platyphylla (Maui). On intergeneric hybrids information. Ifisland age can be used as a rough between Argyroxyphium sandwicense and D. time frame, these genetic distances correspond menziesii (Maui) both parental plant species sup­ to approximately 2% base-pair divergence per ported one planthopper species each and all million years. Although there may be other spe­ apparent hybrids shared the species associated cies not examined here that would fall within with D. menziesii. Herbivorous Insects on Hawaiian Silverswords-RoDERICK 445

A. Silversword Alliance B. Nesosydne Planthoppers

P. marginata ------. .------Calif.Tarweeds N.koae N.naenae D.p. N. naenae 62 K _ N.naenae N. naenae .------D. ptantaginea D.knudsenii ___

D. taxa --- r.N.,--.-ch~a=m:;b=e=rs::;i=' 95 D. imbricata ~ D.c. N. chambersi b L- D. taevigata N. chambersi 80 81 54 b .--- D. sherffiana D.s. IN. chambersi Ic y 100 N. raillardiicota d 100 ...----IJ;; H N. raillardiicota 100 92 1":":"+--- N. raillardiicola H ~ D.m. N. raillardiicota N. raillardiicota 87 M N. raillardiicota N. raillardiicola 2% Seq. Var

100 99

FIGURE 3. Comparison of phylogenies of (A) select species of the Hawaiian silversword alliance (data from Baldwin and Robichaux [1995]) with (B) associated Nesosydne planthoppers. Shaded ovoid areas in A represent plant species that hybridize. Boxes in B represent individuals of Nesosydne species collected on the same host plant noted. Letters a-e denote points of cospeciation as recovered by TreeMap (see text). Bootstrap percentages for each node are shown. The trifurcation near the base of both trees is left unresolved because of low bootstrap support. Nesosydne koae feeds on koa in Hawai'i, and Prokelisia marginatQ is a salt marsh-inhabiting delphacid planthopper from the U.S. mainland (see text).

DISCUSSION may not be important in explaining significant cospeciation. Though still in progress, work to date sug­ Some biological observations can shed light gests that the Nesosydne planthoppers associated on this issue. Research on other delphacid plant­ with the Hawaiian silversword alliance are hoppers suggests that planthoppers can adapt to highly host-specific, with each species feeding on one or a few closely related hosts. The signifi­ closely related novel hosts. For example, numer­ cant pattern of cospeciation for the species ous studies have shown that the rice brown plan­ examined to date suggests that species of plant­ thopper, Nilaparvata lugens, can overcome new hoppers and members ofthe silversword alliance "resistance genes" in rice in only a few genera­ share parallel or cophylogenies and is consistent tions, but that wider host-switching is with reciprocal adaptation. Unfortunately, this limited (for a review see Roderick [1994]). result does not directly demonstrate reciprocal Delphacid planthoppers contain yeastlike endo­ adaptation nor identify the processes that under­ symbionts that may be involved in this adapta­ lie the pattern of cophylogenies (Price 1996). tion, although their role has yet to be identified. For example, cophylogenies may arise through Investigation of hybrid zones presented here a number of processes that may include vicari­ demonstrated that single planthopper species are ance in one or both players, coevolution, or adap­ not found on distantly related hosts, despite sym­ tation by only one player in response to the other. patry and the existence of intermediate hybrids. It is possible that major vicariance events, such In two hybrid zones, planthoppers do not feed on as island and volcano formation, have shaped one ofthe two parental species. These patterns of both planthopper and plant phylogenies concur­ host association indicate that limits exist to host rently, and that planthopper adaptation to hosts adaptation by planthoppers. Also, that single 446 PACIFIC SCIENCE, Volume 5 I, October 1997

Locality of Host Habitat Relationship Host Parent... backcross ... F1 ... backcross ...Parent Hybrid Zone of 2 Host Parents

A. Maui: Haleakala Hybrids and parents overiap distant D. menziesii ------. A. sandwicense m. Crater on slopes and valley floor (inter- I II I generic) N. rail/ardiico/a N. argyroxiphii

B. Maui: Haleakala Hybrids and parents overiap closest D. menziesii ------D. p/afyphylla Outside slope in ravine relatives I I N. bridwelli

C. Big Island: Hybrids and parents overiap closest D. cilio/afa g. ------D. arborea Mauna Kea on slopes and ravine relatives I I N. chambersi

D. Big Island: Parents allopatric: D.c. on close D. cilio/afa g. ------. D. scabra Saddle pahoehoe lava, D.s. and relatives I I hybrid on aa lava N. rail/ardiae

E. Kauai: Alakai Parents allopatric: D.p. on closest D. pa/eafa ------D. raillardioides Swamp bogs, OJ. forest, hybrid relatives I I on forest margins N. naenae

FIGURE 4. Patterns of planthopper host plant use and characteristics of five hybrid zones between members of the Hawaiian silversword alliance (dashed lines). Range of hybrid use for each planthopper species is noted by rectangles. In two hybrid zones, parental plant species were found with no planthoppers. species feed on some closely related hosts but species have been reared from more than one not on others suggests that some closely related species (Hardy and Delfinado 1980). Further, plant species have not diverged sufficiently to some of these species feed on distantly related limit planthopper distribution. These observa­ taxa that are sympatric, indicating at least some tions are consistent with the hypothesis that degree of host-switching. Because some mem­ diversity of Nesosydne planthoppers parallels bers ofthe silversword alliance can go for many and follows the diversity generated in the sil­ years without flowering, using other hosts for versword alliance. Current work on this group flower feeding may be a necessity (although includes reciprocal transplant studies to examine prolonged pupation as found in some North the role of plant hybridization in the diversifica­ American tephritids may also occur [D. Papaj, tion of Nesosydne planthoppers, population pers. comm.]). Whether host-associated geno­ genetic assessments ofplanthoppers across plant types and/or sibling species exist (see Bush hybrid zones, as well as a complete analysis of 1994), awaits further analysis. Other hypotheses all Hawaiian Nesosydne planthoppers including to be tested with this group include whether the species that feed on taxa other than the sil­ tephritids in Hawai'i stem from a single origin, versword alliance. whether gall forming is ancestral to flower feed­ Preliminary observations of patterns of host ing, and whether there has been a single host use by other insect groups associated with the switch between another composite, Bidens, and Hawaiian silversword alliance indicate that pat­ the silversword alliance (Hardy and Delfinado terns of cospeciation are not universal. For 1980; Asquith, Seiler, Miramontes, and Mess­ example, endemic tephritid flies (Diptera: ing, unpubl. data). Molecular work to address Tephritidae) associated with the Hawaiian sil­ these hypotheses is under way. versword alliance are less host-specific, with The insects associated with the Hawaiian sil­ many species feeding in flowers of more than versword alliance provide a unique opportunity one host. For example, only seven of 12 species to examine historical patterns of host plant use have been reared from only one host species and, perhaps more important, the mechanisms within the silversword alliance, whereas five that underlie those patterns. Whether patterns of Herbivorous Insects on Hawaiian Silverswords-RoDERICK 447 cospeciation are a consequence of coevolution and G. D. CARR. 1991. Chloroplast DNA has yet to be resolved and awaits further study. evidence for a North American origin of the Hawaiian silversword alliance (Asteraceae). Proc. Natl. Acad. Sci. U.S.A. 88:1840-1843. ACKNOWLEDGMENTS BOECKLEN, W. J., and R. SPELLENBERG. 1990. Structure of herbivore communities in two I thank U. Bergstrom, A. Isagawa, E. Metz, oak (Quercus spp.) hybrid zones. Oecologia and B. Thorsby for technical help. For assistance (Ber!.) 85:92-100. in collecting, I am grateful to R. Gillespie, R. BROOKS, D. R. 1979. Testing the context and Bartlett (Maui Land and Pineapple), E. Metz, extent of host-parasite coevolution. Syst. and R. Robichaux. B. Baldwin, R. Gillespie, E. Zool. 28:299-307. Metz, S. Palumbi, R. Robichaux, and F. Villa­ ---. 1988. Macroevolutionary comparisons blanca provided helpful discussions and sugges­ of host and parasite phylogenies. Annu. Rev. tions. Comments from R. DeSalle, B. Farrell, R. Eco!. Syst. 19:235-259. Gillespie, J. Liebherr, and D. Polhemus greatly BUSH, G. 1994. Sympatric speciation in animals: improved the manuscript. Finally, I thank J. New wine in old bottles. Trends Eco!. Liebherr and D. Polhemus for organizing the Evol. 9:285-288. Perkins Symposium and the resulting CARR, G. D. 1985. Monograph of the Hawaiian publication. Madiinae (Asteraceae): Argyroxiphium, Dubautia, and Wilkesia. Allertonia 4: 1-123. ---. 1987. Beggar's ticks and tarweeds: LITERATURE CITED Masters of adaptive radiation. Trends Eco!. Evo!. 2:192-195. AGUILAR, J. M., and W. J. BOECKLEN. 1992. ---. 1990a. 11. Asteraceae. 9. Argyroxiph­ Patterns of herbivory in the Quercus grisea ium DC Silversword. Pages 258-262 in W. X Quercus gambelii species complex. L. Wagner, D. R. Herbst, and S. H. Sohmer, Oikos 64:498-504. eds. Manual of the flowering plants of ASCHE, M. 1997. A review of the systematics Hawai'i, Vo!. 1. University of Hawai'i Press, of Hawaiian planthoppers (Hemiptera: Ful­ Bishop Museum Press, Honolulu. goroidea). Pac. Sci. 51:366-376. ---. 1990b. 11. Asteraceae. 26. Dubautia ASQUITH, A. 1995. Evolution of Sarona (Heter­ Gaud. Na'ena'e, kupaoa. Pages 292-308 in optera, Miridae). Pages 90-120 in W. L. W. L. Wagner, D. R. Herbst, and S. H. Wagner and V. A. Funk, eds. Hawaiian bioge­ Sohmer, eds. Manual of the flowering plants ography: Evolution on a hot spot archipelago. of Hawai'i, Vo!. 1. University of Hawai'i Smithsonian Institution Press, Washington. Press, Bishop Museum Press, Honolulu. BALDWIN, B. G. 1997. Adaptive radiation of the CHRISTENSEN, K. M., T. G. WHITHAM, and P. Hawaiian silversword alliance: Congruence KEIM. 1995. Herbivory and tree mortality and conflict of phylogenetic evidence from across a pinyon pine hybrid zone. Oecologia molecular and non-molecular investigations. (Ber!.) 101 :29-36. In T. Givnish and K. J. Sytsma, eds. Molecu­ CLARY, D.O., and D. R. WOLSTENHOLME. 1985. lar evolution and adaptive radiation. Cam­ The mitochondrial DNA molecule of Dro­ bridge University Press, Cambridge. sophila yakuba: Nucleotide sequence, gene BALDWIN, B. G., and R. H. ROBICHAUX. 1995. organization, and genetic code. J. Mol. Historical biogeography and ecology of the Evo!. 22:252-271. Hawaiian silversword alliance (Asteraceae): DENNO, R. F., and G. K. RODERICK. 1990. Popu­ New molecular phylogenetic perspectives. lation biology of planthoppers. Annu. Rev. Pages 259-287 in W. L. Wagner and V. A. Entomo!. 35:489-520. Funk, eds. Hawaiian biogeography: Evolu­ DENNO, R. F., G. K. RODERICK, M. A. tion on a hot spot archipelago. Smithsonian PETERSON, A. F. HUBERTY, H. G. DOBEL, M. Institution Press, Washington. D. EUBANKS, J. E. LOSEY, and G. A. LANGEL­ BALDWIN, B. G., D. W. KYHOS, J. DVORAK, LOTTO. 1996. Habitat persistence and dis- 448 PACIFIC SCIENCE, Volume 51, October 1997

persal in planthoppers. Ecol. Monogr. tions among herbivorous insects and plants. 66:389--408. Pages 207-231 in D. J. Futuyma and M. Slat­ DETHIER, V. G. 1954. Evolution offeeding pref­ kin, eds. Coevolution. Sinauer, Sunderland, erences in phytophagous insects. Evolution Massachusetts. 8:833-854. ---. 1983b. Selective factors in the evolution DOBLER, S., P. MARDULYN, J. M. PASTEELS, of host choice by phytophagous insects. and M. ROWELL-RAHIER. 1996. Host-plant Pages 227-244 in S. Ahmad, ed. Herbivorous switches and the evolution of chemical insects. Academic Press, New York. defense and life history in the leaf beetle GILLESPIE, R. G. 1997. Evolution and extinction genus Oreina. Evolution 50:2373-2386. of Hawai'i's unique biota. HMS Beagle 2:24. EHRLICH, P. R, and P. H. RAVEN. 1964. Butter­ GILLESPIE, R. G., H. B. CROOM, and G. L. flies and plants: A study in coevolution. Evo­ HASTY. 1997. Phylogenetic relationships and lution 18:586-608. adaptive shifts among major clades of Tetra­ ELDREDGE, L. G., and S. E. MILLER. 1995. How gnatha spiders (Araneae: Tetragnathidae) in many species are there in Hawaii? Bishop Hawai'i. Pac. Sci. 51:380-394. Mus. Occas. Pap. 41:3-18. HAFNER, M. S., P. D. SUDMAN, F. X. VILLA­ FARRELL, B. D., and C. MITTER. 1993. Phyloge­ BLANCA, T. A. SPRADLING, 1. W. DEMASTES, netic determinants of insect/plant community and S. A. NADLER. 1994. Disparate rates of diversity in R. E. Ricklefs and D. Schluter, molecular evolution in cospeciating hosts and eds. Species diversity in ecological communi­ parasites. Science (Washington, D.C.) ties. University of Chicago Press, Chicago. 265:1087-1090. FLOATE, K. D., and T. G. WHITHAM. 1993. The HARDY, D. E., and M. D. DELFINADO. 1980. "hybrid bridge" hypothesis: Host shifting via Diptera: Cyclorrhapha III, Series Schizo­ plant hybrid swarms. Am. Nat. 141:651-662. phora Section Acalypterae, exclusive ofFam­ ---. 1994. Aphid-ant interaction reduces ily Drosophilidae. Insects of Hawaii. chrysomelid herbivory in a cottonwood University Press of Hawai'i, Honolulu. hybrid zone. Oecologia (Berl.) 97:215-221. HOWARTH, F. G., and W. P. MULL. 1992. Hawai­ FLOATE, K. D., M. 1. C. KEARSLEY, and T. G. ian insects and their kin. University of WHITHAM. 1993. Elevated herbivory in plant Hawai'i Press, Honolulu. hybrid zones: Chrysomela confluens, popu­ KIMURA, M. 1980. A simple method for estimat­ lus, and phenological sinks. Ecology ing evolutionary rate of base substitutions 74:2056-2065. through comparative studies of nucleotide FRIAR, E. A., R H. ROBICHAUX, and D. W. sequences. J. Mol. Evol. 16:111-120. MOUNT. 1996. Molecular genetic diversity LIEBHERR, J. K., and D. A. POLHEMUS. 1997. following a population crash in the endan­ Comparisons to the century before: The leg­ geredMaunaKea silversword,Argyroxiphium acy of R C. L. Perkins and Fauna Hawai­ sandwicense spp. sandwicense (Asteraceae). iensis as the basis for a long-term ecological Mol. Ecol. 5:687-691. monitoring program. Pac. Sci. 51 :490-504. FRITZ, R S., C. M. NICHOLS-ORIANS, and S. J. MADDISON, W. P., and D. R MADDISON. 1992. BRUNSFELD. 1994. Interspecific hybridiza­ MacClade: Analysis of phylogeny and char­ tion of plants and resistance to herbivores: acter evolution. Sinauer, Sunderland, hypotheses, genetics, and variable responses Massachusetts. in a diverse herbivore community. Oecologia MILLER, S. E., andL. G. ELDREDGE. 1996. Num­ (Berl.). 97:106-117. bers of Hawaiian species: Supplement 1. FUNK, D. J., D. J. FUTUYMA, G. ORT!, and A. Bishop Mus. Occas. Pap. 45:8-17. MEYER. 1995. A history of host associations MITTER, c., and D. R. BROOKS. 1983. Phyloge­ and evolutionary diversification for Ophrella netic aspects of co-evolution. Pages 65-98 in (Coleoptera: Chrysomelidae): New evidence D. J. Futuyma and M. Slatkin, eds. Coevolu­ from mitochondrial DNA. Evolution tion. Sinauer, Sunderland, Massachusetts. 49:1008-1017. MITTER, C., B. FARRELL, and D. J. FUTUYMA. FUTUYMA, D. 1. 1983a. Evolutionary interac- 1991. Phylogenetic studies of insect-plant Herbivorous Insects on Hawaiian Silverswords-RoDERlCK 449

interactions: Insights into the genesis of mitochondrial gene sequences and a compila­ diversity. Trends Ecol. Evol. 6:290-293. tion of conserved polymerase chain reaction MITIER, C., B. FARRELL, and B. WIEGMANN. primers. Ann. Entomo!. Soc. Am. 87: 1988. The phylogenetic study of adaptive 651-701. zones: Has phytophagy promoted insect STRAUSS, S. 1994. Levels ofherbivory and para­ diversification? Am. Nat. 132:107-128. sitism in host hybrid zones. Trends Eco!. MORAN, N., and P. BAUMANN. 1994. Phyloge­ Evo!. 9:209-214. netics of cytoplasmically inherited microor­ SWEZEY, O. H. 1954. Forest entomology in ganisms of arthropods. Trends Ecol. Evol. Hawaii. Bishop Museum, Honolulu. 9:15-20. SWOFFORD, D. L. 1993. PAUP. Phylogenetic NISHIDA, G., ed. 1994. Hawaiian terrestrial analysis using parsimony. Sinauer, Sunder­ arthropod checklist. Hawai'i Biological Sur­ land, Massachusetts. vey. Bishop Museum Press, Honolulu. TAMURA, K, and M. NEI. 1993. Estimation of O'CONNELL, C. 1991. Acoustic communication the number of nucleotide substitutions in the and sexual selection in a group of Hawaiian control region of mitochondrial DNA in planthoppers (Homoptera: Delphacidae). humans and chimpanzees. Mol. Bio!. Evol. Master's thesis, University of Hawai'i at 6:447-459. Manoa, Honolulu. THOMPSON, 1. N. 1994. The coevolutionary pro­ PAGE, R. D. M. 1995a. Parallel phylogenies: cess. University of Chicago Press, Chicago. Reconstructing the history of host-parasite WAGNER, W. L., and V. A. FUNK, eds. 1995. assemblages. Cladistics 10: 155-173. Hawaiian biogeography: Evolution on a hot 1995b. TreeMap. University of spot archipelago. Smithsonian Institution Oxford, Oxford. Press, Washington. PRICE, P. W. 1996. Biological evolution. Saun­ WHITHAM, T. G., P. A. MORROW, and B. M. ders College Publishing, Fort Worth, Texas. POTIS. 1994. Plant hybrid zones as centers RODERICK, G. K 1987. Ecology and evolution of biodiversity: The herbivore community of of dispersal of a pacific salt marsh insect. two endemic Tasmanian eucalypts. Oeco­ Ph.D. diss., University of California, logia (Ber!.) 97:481--490. Berkeley. WIEGMANN, B. M., C. MITTER, and B. FARRELL. ---. 1994. Genetics of host plant adaptation 1993. Diversification ofcarnivorous parasitic in delphacid planthoppers. Pages 551-570 in insects: Extraordinary radiation or special­ R. F. Denno and T. 1. Perfect, eds. Planthop­ ized dead end. Am. Nat. 142:737-754. pers: Their ecology and management. Chap­ WILSON, E. O. 1996. Insects: The ultimately man and Hall, London. biodiverse animals. Page xvi in Proceedings, RODERICK, G. K and R. G. GILLESPIE. Specia­ XX International Congress of Entomology, tion and phylogeography of Hawaiian terres­ Firenze, Italy. trial arthropods. Mo!. Eco!. in press. ---, ed. 1988. Biodiversity. National Acad­ RODERICK, G. K, and E. C. METZ.1997. Biodiv­ emy Press, Washington. ersity of planthoppers (Hemiptera: Delphaci­ WILSON, W. W., C. MITTER, R. F. DENNO, and dae) on the Hawaiian silversword alliance: M. R. WILSON. 1994. Evolutionary patterns Effects of host plant history and hybridiza­ of host plant use by delphacid planthoppers tion. Mem. Mus. Vic. 56:393-399. and their relatives. Pages 7-113 in R. F. SIMON, C. 1987. Hawaiian evolutionary biology: Denno and T. 1. Perfect, eds. Planthoppers: An introduction. Trends Eco!. Evo!. Their ecology and management. Chapman 2:175-178. and Hall, New York. SIMON, c., F. FRATII, A. BECKENBACH, B. ZIMMERMAN, E. C. 1948. Homoptera: Auche­ CRESPI, H. Lru, and P. FLOOK. 1994. Evolu­ norhyncha. Insects of Hawaii. University of tion, weighting, and phylogenetic utility of Hawai'i Press, Honolulu.