Proc. Nadl. Acad. Sci. USA Vol. 82, pp. 4753-4757, July 1985 Ancient origin for Hawaiian inferred from protein comparisons (/Scptomyza/larval hemolymph protein/microcomplement fixation/Hawaiian geology) STEPHEN M. BEVERLEY*t AND ALLAN C. WILSON* *Department of Biochemistry, , Berkeley, CA 94720; and tDepartment of Pharmacology, Harvard Medical School, Boston, MA 02115 Communicated by Hampton L. Carson, March 25, 1985

ABSTRACT Immunological comparisons of a larval we recently showed that this may apply to LHPs in more than hemolymph protein enabled us to build a tree relating major 30 species of Drosophila and related , including two groups of drosophiline flies in Hawail to one another and to lineages of (11). The conclusion was continental flies. The tree agrees in topology with that based on that the variance in rate of LHP evolution is low enough to internal anatomy. Relative rate tests suggest that evolution of permit the use of LHP as a tool for estimating times of hemolymph proteins has been about as fast in Hawaii as on divergence (11). continents. Since the absolute rate of evolution of bemolymph This report extends our studies to 18 species of Hawaiian proteins in continental flies is known, one can erect an drosophilines, including members of the . approximate time scale for Hawaiian evolution. According Our analysis suggests that rates of LHP evolution are not to this scale, the Hawaiian fly fauna stems from a colonist that accelerated within the Hawaiian drosophilines, supporting landed on the archipelago about 42 million years ago-i.e., the use of LHP as an estimator of divergence times. The before any of the present islands harboring drosophilines divergence times estimated from the LHP distances indicate formed. This date fits with the geological history of the that an ancient origin model is the correct one for the archipelago, which has witnessed the sequential rise and Hawaiian drosophilines. erosion of many islands during the past 70 million years. We discuss the bearing of the molecular time scale on views about METHODS AND MATERIALS rates of organismal evolution in the Hawaiian flies. Fly Strains and LHP Extracts. Of the 21 species examined The Hawaiian archipelago has received extensive study as (Table 1), 3 are continental and 18 are Hawaiian; included are the setting of seemingly rapid biological evolution within representatives of two genera, Drosophila and Scaptomyza. many taxonomic groups (1, 2). Hawaiian members of the Third instar larvae of these species were provided by the subfamily Drosophilinae, for example, comprise more than National Drosophila Species Stock Center (Univ. of Texas, 400 described species representing more than 15% of the Austin) or by Herman Spieth (Univ. of California at Davis). world's drosophilines, yet Hawaii accounts for less than Protein extracts containing monomeric LHP were made from 0.01% of the world land area (3-5). This observation could third instar larvae as described (10). imply that evolution in Hawaii has been extremely fast, ifthis Protein Comparisons. Immunological distances between large group offly species were no older than the oldest island pairs of LHPs were measured by the quantitative Kauai, or about 5-6 million years (Myr) (6). Another pos- microcomplement fixation method with antisera to highly sibility is that the fly fauna is far older (3, 4, 7, 8). Even though purified LHPs. The five antisera used and the way of doing the fossil record of Hawaiian flies is poor, much is known microcomplement fixation tests with LHPs have been de- about the geological history of the archipelago and the scribed (10). All immunological distances were checked phylogenetic relationships of some of these flies. For one qualitatively by the Ouchterlony double-diffusion test (10). subgroup, in particular, it has been possible through intensive Immunological distance (y) between 290 pairs of monomeric cytogenetic, biogeographic, and radiometric dating studies to proteins is correlated (r = 0.9) to the percentage difference in develop an evolutionary time scale; we refer to the picture- amino acid sequence (x) by an equation ofthe formy = Sx (10, winged flies of the planitibia subgroup, which appears to be 24). The average error in measuring an immunological dis- less than 5 Myr old (3, 9). tance between two LHPs (e.g., A and B) with one antiserum Our approach has been to build a temporal framework for (e.g., anti-A) is ±2.6 units (10), and the average deviation the evolution of these flies by using biochemical methods of from perfect reciprocity (i.e., anti-A vs. B compared with estimating approximate times ofdivergence between lineages anti-B vs. A) is 14.2% of the mean distance (10). leading to living species. We use immunological comparisons of a larval hemolymph protein (LHP) found throughout the RESULTS AND DISCUSSION higher diptera to estimate degree of amino acid sequence Immunological Distances. The immunological distances divergence (10, 11), an approach used with success in the observed between the LHPs of the 5 reference species and study of vertebrate proteins and evolution (12). Attesting to those of 16 additional Hawaiian species are presented in the validity of this approach is the fact that the phylogenetic Table 1. The antiserum to Drosophila crucigera LHP sepa- tree constructed from the immunological distances observed rates the 9 species tested within the picture-winged group into among drosophiline LHPs agrees approximately in branching two divisions. Division I has LHPs that are nearly identical order with that based on 60 anatomical traits ofdrosophilines with D. crucigera LHP, the immunological distances from (13, 14). Point mutations are now known to accumulate at this reference species being 4 or less. Included in this division fairly steady rates in the proteins of vertebrates (12, 15) and are LHPs from 7 species representing most of the recognized subgroups within the picture-winged group (5, 8). Division II The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" Abbreviations: LHP, larval hemolymph protein; Myr, millions of in accordance with 18 U.S.C. §1734 solely to indicate this fact. years. 4753 Downloaded by guest on October 2, 2021 4754 Evolution: Beverley and Wilson Proc. Nad Acad Sd USA 82 (1985) Table 1. Immunological distances within Hawaiian Drosophilinae Table 1 and refs. 10 and 11. This tree (Fig. 1), which is in Immunological distance general agreement with that based on anatomy and behavior (4, 25, 26), depicts two clusters of lineages, one leading from Source of LHP Cru Mim Sca Mul Mel node E to the LHPs of 14 Hawaiian Drosophila species and Genus Drosophila the other leading to the LHPs offour Scaptomyza species and Subgenus the subgenus Engiscaptomyza. The remaining two lineages, D. melanogaster (C)* 76 90 80 74 0 leading from node D to continental species of the subgenus Subgenus Drosophila Drosophila and from node C to the subgenus Sophophora D. mulleri (C) 50 59 56 0 78 signify the phylogenetic position occupied by the LHPs of Picture-winged group Drosophila mulleri and Drosophila melanogaster, respec- LHP division I tively, as well as other continental species studied (11). D. crucigera 0 32 53 58 84 The thickest bar in the protein tree is intended to point out D. conspicua -3 NP NP NP NP that nodes D (54.3 + i.9; results represent mean + SEM) and D. odontophallus NPt 34 NP NP NP E (52.9 ± 3.4) are not clearly resolved, possibly because the D. orthofascia -2 NP N? NP NP immunological distance method provides only an approxi- D. picticornis 4 NP NP NP NP mate estimate of the extent of protein divergence. For this D. punalua 1 NP NP NP NP reason, one cannot rule out the possibility that certain D. silvarentis 2 NP NP NP NP continental species-specifically, those belonging to the LHP division II pinicola, immigrans, robusta, and virilis groups and the D. adiastola 34 49 37 49 NP subgenus * (see figure 5 of ref. 11)-are as D. setosimentum 40 53 NP 49 NP closely related to Scaptomyza and Hawaiian Drosophila as Modified-mouthparts group those two groups are to each other. Hence, the possibility D. mimica 17 0 54 52 79 that the Hawaiian archipelago was colonized twice at about D. biseriata NP 25 NP NP NP the same time, once by the ancestor ofScaptomyza and once D. dissita 39 13 75 NP NP by the ancestor ofthe Drosophila, cannot be ruled out. On the D. eurypeza 36 26 80 NP NP grounds of parsimony, as well as Throckmorton's (4) obser- D. hirtitarsus 23 19 58 NP NP vations on the morphological traits of Hawaiian Drosophila Subgenus Engiscaptomyzat and Scaptomyza, the hypothesis of a single colonization D. amplilobus 39 44 25 61 61 event followed by immediate divergence is favored. Genus Scaptomyza Adiastola Subgroup (Division H). The tree analysis shows Subgenus Parascaptomyza that the LHPs of picture-winged flies consist of two S. adusta (C) 42 57 0 57 70 phylogenetically distinct groups: division I (most species) S. elmoi 33 59 24 61 NP and division II (the adiastola subgroup; Fig. 1). Consistent Subgenus Bunostoma with this finding, the mean LHP distance between the S. varifrons 42 71 40 NP 69 division I and the modified-mouthparts groups (node G; 30.2 Subgenus Trogloscaptomyza + 3.5) is significantly smaller than that between division II S. affiniscuspidata 47 55 29 61 NP and the former two groups (node F; 44 ± 4.3). To check on the possibility that the large LHP distance Immunological distances were determined by reaction with anti- sera to LHPs from various species. Cru, anti-D. crucigera; Mim, between the two divisions is due to accelerated evolution on anti-D. mimica; Sca, anti-S. adusta; Mul, anti-D. mulleri; Mel, the lineage leading to division II, we applied the concept of anti-D. melanogaster. NP, comparison not performed. the relative rate test (10-12). The observation that the mean *C, continental species included in this table for reference. distance from division II to Scaptomyza and D. mulleri (43 tPlaced in LHP division I by immunodiffusion test. units) is actually less than that from division I to Scaptomyza tThis subgenus, comprising the crassifemur and nasalis subgroups, and D. mulleri (55.5 units) rules out this possibility. Accord- has traditionally been classified in the genus Drosophila (16, 17). In ingly, the large LHP distance between the division I and II most organismal respects, however, this subgenus has more in flies implies an ancient divergence. common with Scaptomyza than with Drosophila [e.g., internal Quantitative support for this inference comes from a anatomy (4), structure of the external genitalia and phallic organ (18-20), and behavior (21)]. This is also the case for the karyotype statistical test using the F parameter (27), which evaluates the (22, 23) and the present LHP data. agreement between the observed LHP distances and the reconstructed distances from the tree. For the branching order shown in Fig. 1, the goodness of fit is much better (F consists of LHPs of Drosophila adiastola and Drosophila = 4%) than for the tree that links divisions I and II (F= 22%) setosimentum, having LHPs that differ from crucigera LHP or the tree that makes a three-way split among divisions I and by 34-40 units of immunological distance-i.e.,1by nearly as II and the modified-mouthparts group (F = 34%). much as do the LHPs ofother groups ofHawaiian flies. Both The proposal that division I flies are more closely related of these flies are members of the adiastola subgroup of to the modified-mouthparts group than either is to the picture-winged flies. adiastola subgroup (division II) disagrees with two previous The antiserum to Drosophila mimica LHP reacts best with views about phylogenetic relationships among these species. LHPs from other flies of the modified-mouthparts group One view, which links division I and II flies (5), points to the (13-26 units) and better with the LHPs of flies in division I chromosomal resemblance between these two types offlies. (32-34 units) than with the LHPs ofthe division II flies (49-53 Since the chromosomes of other groups of flies, including units). The antiserum to Scaptomyza adusta LHP reacts best those ofthe modified-mouthparts group, differ so much from with the LHPs of Scaptomyza species (all Hawaiian) and the those of the picture-winged flies, an accurate phylogenetic Hawaiian subgenus Engiscaptomyza, which is traditionally analysis ofthe inversion relationships and evolutionary rates classified in the genus Drosophila (5, 16). Table 1 also shows among these types of fly karyotypes has not been presented that the LHPs of the Hawaiian flies are generally more antigenically similar to one another than to continental tThese continental groups are among those frequently cited as Drosophila LHPs. potential "closest" relatives of the Hawaiian flies; the LHP data Protein Tree. A molecular tree summarizing the immuno- (refs. 10 and 11 and Fig. 1) suggest that they are all approximately logical relationships was constructed (10) from the data of equally related. Downloaded by guest on October 2, 2021 Evolution: Beverley and Wilson Proc. NatL Acadl Sci USA 82 (1985) 4755 Koko Seamount 0 \ Midway 2z ocz.ICc$~ Hawaii

Picture-winged (I)

} Modified mouth parts

adiastola (I:)

Parascaptomyza Trog loscaptomyza Engiscaptomyza

Bunostoma

Continental Drosophila

Sophophora

60 20 0 Time (Myr) - I 75 50 25 0 Immunological Distance

FIG. 1. Molecular tree for Hawaiian flies in relation to time and geography. (Lower) Immunological relationships among the LHPs of the Hawaiian fly species listed in Table 1. Continental Drosophila refers to the Hirtodrosophila, immigrans, pinicola, repleta, virilis, and robusta groups; Sophophora includes members of the melanogaster, obscura, willistoni, and saltans groups (see refs. 10 and 11). The topology of the tree was established by the method of Beverley and Wilson (10). Thick vertical bars signify nodes leading to multiple lineages among which the exact relationships are ambiguous. The position of each node corresponds to the average of unidirectional immunological distances between species connected through that node; these values are D, 54.3 + 1.9 (n = 30); E, 52.9 3.4 (n = 16); F, 44 + 4.3 (n = 4); G, 30.2 + 3.4 (n = 6). The time scale above the immunological distance scale is based on the immunological distance scale as described in the text. (Upper) Bathymetric outline at a depth of 3600 m (1) for the Hawaiian archipelago, plotted so that the position of each island (dark regions) and undersea platform corresponds approximately to their geologic age.

(cf. refs. 28-30). A second view links division II with the Model 1. A recent origin. The Hawaiian flies are actually modified-mouthparts flies and is supported by some morpho- no more than 5-6 Myr old, and the LHP-derived time is logical and behavioral evidence (8). The view corresponding inaccurate. One explanation would be that the rate of LHP to the LHP tree in Fig. 1 receives support from studies of evolution has been accelerated in Hawaiian drosophilines, ovarian transplantation (31). There has been too little elec- leading to an inaccurate molecular estimate (an overestimate) trophoretic work on proteins (32, 33) or DNA hybridization of the time of origin of the Hawaiian fauna on Kauai. (17, 34, 35) to have a significant bearing on the problem ofthe Model 2. An ancient origin. The Hawaiian flies originated phylogenetic position of the adiastola subgroup. Further about 40 Myr ago, and the rate of protein evolution is the work is required to establish the position of this subgroup same in both continental and Hawaiian species. In this among Hawaiian flies. model, the flies are thought to have colonized islands in the Subgenus Engiscaptomyza. The LHP tree clearly links geographical vicinity of the present islands. Drosophila amplilobus (subgenus Engiscaptomyza) with the It is first necessary to consider whether the 42-Myr Hawaiian Scaptomyza, and not with the Hawaiian estimate from LHP comparisons is significantly different Drosophila. The strong phenotypic resemblance and close from the 6-Myr age of Kauai. By considering the mean LHP phylogenetic association between the Engiscaptomyza and distance for the oldest divergence within the Hawaiian flies Scaptomyza has been recognized by others (4, 16, 18-23, 25). (node E in Fig. 1, 53 3.4 units) and the range in the LHP Molecular Time Scale. The time scale in Fig. 1 comes from evolutionary rate (1.13-1.39 units/Myr; ref. 11), we find the the calibrated rate ofevolution ofLHPs in Drosophila and the 95% confidence limits for the time of origin of the Hawaiian higher Diptera, 1.25 immunological distance units per Myr flies to be 33-53 Myr, significantly different than 6 Myr. since divergence (11). It appears that the colonization of the Relative Rate Tests. The steadiness of LHP evolution was Hawaiian islands took place about 42 Myr ago-i.e., long assessed by using relative rate tests (10-12). Using the tree before the oldest present-day island harboring drosophilines topology shown in Fig. 1, we calculated the average distances (Kauai) formed 5-6 Myr ago (6). We consider two alternative from the common ancestral nodes C and D to the modern models in addressing this discrepancy: Hawaiian and continental species of LHP. The ratio of the Downloaded by guest on October 2, 2021 4756 Evolution: Beverley and Wilson Proc. NatL Acad ScL USA 82 (1985) Hawaiian distance to the continental distance is 0.78 for node Fig. 1 juxtaposes the phylogenetic tree for Hawaiian C and 1.16 for node D, the mean value being 0.97. Thus, drosophilines and a map of this part of the Pacific Ocean, so significant acceleration in LHP evolution in the Hawaiian that the time scale for the tree corresponds approximately to lineages required by a recent origin model is not observed. the time of formation of the islands and seamounts of the Another approach is to superimpose the immunological Hawaiian chain. The inference is that colonization of the data on the recent-origin model and then examine the archipelago by the founding species ofDrosophila took place uniformity ofthe rate of LHP evolution (Fig. 2). This scheme shortly after formation of the island corresponding to the would require 7-fold-accelerated LHP evolution on the Ha- Koko seamount, now lying 3000 km northwest ofHawaii. As waiian islands. However, this model also requires a long Koko Island eroded away, younger islands emerged in period of anomalously slow LHP evolution (0.3 of the succession to the southeast and were presumably populated standard rate) along the lineage leading to the common by the descendants of flies from Koko Island. We propose ancestor of the Hawaiian and continental drosophilines. As that in a continuing series ofisland-hopping colonizations the there is no available evidence supporting a 20-fold shift in the descendants of these flies eventually colonized the present rate of LHP evolution, we suggest that the recent-origin high islands of Hawaii. model leads to an about rates ofLHP Before our proposal can be accepted, it must satisfy unacceptable prediction several criteria: first, that the islands formed the last evolution. This finding is unaffected by uncertainty concern- during the 40 Myr provided successive targets for colonization. As ing number of independent colonizations. shown in Fig. 1, the seamounts of the archipelago occur at Geological Setting. One can reconcile an ancient-origin intervals comparable with those among the present Hawaiian model with geological information concerning the history of islands. Second, the flies must have been capable of the Hawaiian archipelago (3, 4, 7, 8, 36, 37). Stretching for interisland colonization. The immediate ancestors of the 3000 km to the northwest of the present Hawaiian Islands modern Hawaiian drosophilines are known to have success- there is a chain of coral atolls and reefs (Fig. 1) that are the fully colonized among the modem islands, as shown by surface signs of a series ofundersea mountains and plateaus. studies of polytene chromosomes (3, 5). Third, the ancient These are the vestiges of volcanic islands that formed over a islands must have offered a suitable habitat for drosophilines. hot spot in the earth's mantle near where the present island Extrapolating along bathymetric transects of the undersea of Hawaii is now. Due to the northwesterly movement of the mounts, we can obtain an estimate of their previous height. part of the ocean floor that overlies the hot spot, each new By performing this analysis on extant islands one can obtain volcano is slowly displaced from the source of the magma. an idea of the accuracy of this approach. Our analysis (28) After some millions ofyears, erosion converts these volcanos indicates that many of the seamounts were once comparable to submarine mountains. Consistent with this scenario are the in elevation to Kauai (1500 m), including Gardner Pinnacles potassium/argon dates for the series of islands and undersea (2100 m) and Ojin Seamount (1700 m) and thus are likely to mountains stretching from the big island of Hawaii, which is provide a suitable habitat for drosophilines.§ Other workers less than 1 Myr old, to the Koko seamount, which is 46 Myr have also considered the ancient, now-eroded islands as old (38). Furthermore (and not shown in the figure), the series potentially suitable for flies (3, 4, 8, 40). Hence, the ancient- continues as the Emperor chain until the surface is subducted origin model appears to be fully consistent with available into an oceanic coast molecular, biological, and geological data. trench off the of northeast Asia. The for Other Taxa. For at least 70 time of formation of these remnants Implications Myr there has oldest of the Hawaiian been a succession of islands in the central Pacific that were archipelago has been measured at about 70 Myr (39). capable of supporting life like that found in Hawaii today. One testable prediction of our model is that the introduction Ancient Origin Hawaiian of other species into Hawaii may have occurred during this interval, with a unique colonization date characterizing each group. It would therefore be interesting to apply molecular methods for dating the origin of these groups on Hawaii, Continental especially plants, some groups ofwhich must have preceded the drosophiline fauna and may have persisted to this day. It is already evident from comparative studies on bird proteins (unpublished data) and DNA (41) that the common ancestor SoDhoDhora of the Hawaiian honey creepers (Drepanididae)-the most diverse endemic group of land birds in Hawaii-may have colonized the Hawaiian archipelago as long as 20 Myr ago. Thus, molecular analysis of at least one other group has also Recent Origin provided evidence in accordance with predictions of the Hawaiian ancient origin model. Implications for the Hawaiian Drosophilinae. While an ancient origin for major elements of the Hawaiian flora and Continental fauna could require revision of views about their tempo and mode of evolution, we emphasize that the short time scale proposed here for division I ofthe picture-winged Drosophila is consistent with previous views that, at the supramolecular Sophophora level, this group's evolution has been rapid compared with that ofcontinental Drosophila (3,5, 9, 28). Our findings invite FIG. 2. Predicted rates of LHP evolution according to two models. The distances consideration of what forces may be responsible for a higher immunological shown in Fig. 1 were used to rate of calculate the rate of LHP evolution (normalized to the rate obtained evolution in this specific group of Hawaiian flies. for the Sophophora-Drosophila divergence) assuming a 42-Myr Among the biological properties possibly peculiar to this (ancient-origin model) or a 6-Myr (recent-origin model) time for the separation of the Hawaiian and continental drosophilines. The §Our heights are considered minimal estimates, as we have not divergence of the subgenus Sophophora from these lineages was corrected for seafloor subsidence (38), which would increase the taken as 62 Myr in both models (11). estimated height. Downloaded by guest on October 2, 2021 Evolution: Beverley and Wilson Proc. NatL Acad ScL USA 82 (1985) 4757 group, the following may be relevant: (t) genetic systems 16. Kaneshiro, K. Y. (1969) Univ. Tex. Publ. 6918, 79-84. capable ofrapidly generating and incorporating evolutionary 17. Triantaphyllidis, C. D. & Richardson, R. H. (1982) Genetica change, such as shifts in gene regulation (28, 42, 43); 57, 225. (ii) 18. Hardy, D. E. (1965) ofHawaii (Univ. of Hawaii Press, unusual and complex social behavior (3, 8, 25, 44); and (ii!) Honolulu, HI), Vol. 12. susceptibility to being driven evolutionarily by predation, 19. Takada, H. (1966) Univ. Tex. Publ. 6615, 315-333. particularly by the rapidly evolving Hawaiian birds (3, 25, 44; 20. Hardy, D. E. (1966) Univ. Tex. Publ. 6615, 195-244. cf. ref. 45). This latter view is attractive in that these birds are 21. Spieth, H. T. (1966) Univ. Tex. Publ. 6615, 245-313. also suspected ofaccelerating the evolution ofsuch Hawaiian 22. Clayton, F. E. (1968) Univ. Tex. Publ. 6818, 263-278. plants as the lobeliads on which many picture-winged flies 23. Clayton, F. E. (1969) Univ. Tex. Publ. 6918, 95-110. depend (46). Finally, superimposed on the ofthe flies 24. Benjamin, D. C., Berzofsky, J. A., East, I. J., Gurd, F. R. N., is the rapidly changing Hawaiian environment, attributable to Hannum, C., Leach, S. J., Margoliash, E., Michael, J. G., the frequent appearance and disappearance of islands with Miller, A., Prager, E. M., Reichlin, M., Sercarz, E. E., Smith- Gill, S. J., Todd, P. E. & Wilson, A. C. (1984) Annu. Rev. numerous diverse habitats. A fuller discussion ofthe rates of Immunol. 2, 67-101. evolution at the supramolecular level in Hawaiian flies, 25. Spieth, H. T. (1966) Univ. Tex. Publ. 6615, 245-313. relative to the time scale developed in this paper, will be 26. Throckmorton, L. H. (1975) in Handbook of Genetics, ed. presented elsewhere. King, R. C. (Plenum, New York), Vol. 3, pp. 421-469. 27. Prager, E. M. & Wilson, A. C. (1978) J. Mol. Evol. 11, We thank H. Carson, D. Dobson, D. Hardy, K. Kaneshiro, S. 129-142. Palumbi, H. Ochman, E. Prager, H. Spieth, R. Richardson, V. 28. Beverley, S. M. (1979) Dissertation (Univ. of California, Sarich, D. Schulze, A. Templeton, L. Throckmorton, M. R. Wheel- Berkeley). er, and E. Zimmer for discussions and specimens; anonymous 29. Wilson, A. C., White, T. J., Carlson, S. S. & Cherry, L. M. reviewers for helpful criticisms; and G. B. Dalrymple, who sug- (1977) in Human Cytogenetics: ICN-UCLA Symposia on Mo- gested the method of estimating the heights of seamounts of the lecular and Cellular Biology, eds. Sparkes, R. S., Comings, Hawaiian archipelago. We acknowledge support by National Insti- D. E. & Fox, C. F. (Academic, New York), Vol. 7, pp. tutes of Health Grant GM21509 and National Science Foundation 375-393. Grant DEB81-12412. 30. Bush, G. L., Case, S. M., Wilson, A. C. & Patton, J. L. (1977) Proc. Natl. Acad. Sci. USA 74, 3942-3946. 1. Zimmerman, E. C. (1948) Insects ofHawaii (Univ. of Hawaii 31. Kambysellis, M. P. (1970) J. Exp. Zool. 175, 169-180. Press, Honolulu, HI), Vol. 1. 32. Ayala, F. J. (1975) Evol. Biol. 8, 1-78. 2. Cariquist, S. (1974) Island Biology (Columbia Univ. Press, 33. Johnson, W. E., Carson, H. L., Kaneshiro, K. Y., Steiner, New York). W. M. & Cooper, M. M. (1976) in Isozymes, ed. Markert, 3. Carson, H. L., Hardy, D. E., Spieth, H. T. & Stone, W. S. C. L. (Academic, New York), Vol. 4, pp. 563-584. (1970) in Essays in Evolution and Genetics in Honor of 34. Hunt, J. A., Hall, T. J. & Britten, R. J. (1981) J. Mol. Evol. 17, , eds. Hecht, M. K. & Steere, W. C. 365-367. (Appleton-Century-Crofts, New York), pp. 437-543. 35. Hunt, J. A. & Carson, H. L. (1983) Genetics 104, 353-364. 4. Throckmorton, L. H. (1966) Univ Tex. Publ. 6615, 335-396. 36. Wilson, J. T. (1963) Can. J. Phys. 41, 863-870. 5. Carson, H. L. & Kaneshiro, K. Y. (1976) Annu. Rev. Ecol. 37. Dalrymple, G. B., Silver, E. A. & Jackson, E. D. (1973) Am. Syst. 7, 311-346. Sci. 61, 294-308. 6. McDougall, I. (1964) Geol. Soc. Am. Bull. 75, 107-128. 38. Clague, D. A. & Dalrymple, G. B. (1973) Earth Planet Sci. 7. Beverley, S. M. (1980) Abstracts of the Second International Lett. 17, 411-415. Congress of Systematic and (Univ. of 39. Scholl, D. W., Creager, J. S., Boyce, R. E., Echols, R. J., British Columbia, Vancouver, Canada), p. 112. Fullam, T. J., Grow, J. A., Koizumi, I., Lee, H., Ling, H.-Y., 8. Spieth, H. T. (1982) Evol. Biol. 14, 351-437. Supko, P. R., Stewart, R. J. & Worsley, T. R. (1971) 9. Carson, H. L. (1976) Nature (London) 259, 395-396. Geotimes 16, 12-15. 10. Beverley, S. M. & Wilson, A. C. (1982) J. Mol. Evol. 18, 40. Carson, H. L. (1983) Genetics 103, 465-482. 251-264. 41. Sibley, C. G. & Ahlquist, J. E. (1982) Auk 99, 130-140. 11. Beverley, S. M. & Wilson, A. C. (1984) J. Mol. Evol. 21, 1-13. 42. Templeton, A. R. (1981) Genetics 94, 1011-1038. 12. Wilson, A. C., Carlson, S. S. & White, T. J. (1977) Annu. Rev. 43. Carson, H. L. (1975) Am. Nat. 109, 83-92. Biochem. 46, 573-639. 44. Spieth, H. T. (1974) in Genetic Mechanisms of Speciation in 13. Throckmorton, L. H. (1962) Univ. Tex. Publ. 6205, 207-343. Insects, ed. White, M. J. D. (Reidel, Boston), pp. 94-101. 14. Throckmorton, L. H. (1968) Syst. Zool. 17, 355-387. 45. Wyles, J. S., Kunkel, J. G., Wilson, A. C. (1983) Proc. Natl. 15. Carlson, S. S., Wilson, A. C. & Maxson, R. D. (1978) Science Acad. Sci. USA 80, 4394-4397. 200, 1183-1185. 46. Spieth, H. T. (1966) Am. Nat. 100, 470-473. Downloaded by guest on October 2, 2021