sign in sign up
<<
Home , Carduelinae, Ciridops, Drepanis, Finch, Hawaiian honeycreeper, Himatione

THE RELATIONSHIPS OF THE HAWAIIAN HONEYCREEPERS (DREPANININI) AS INDICATED BY DNA-DNA HYBRIDIZATION

CH^RrES G. SIBLEY AND Jo• E. AHLQUIST Departmentof Biologyand PeabodyMuseum of Natural History, Yale University, New Haven, Connecticut 06511 USA

ABSTRACT.--Twenty-twospecies of Hawaiian honeycreepers(Fringillidae: : Drepaninini) are known. Their relationshipsto other groups of passefineswere examined by comparing the single-copyDNA sequencesof the Apapane (Himationesanguinea) with those of 5 speciesof carduelinefinches, 1 speciesof , 15 speciesof New World nine- primaried oscines(Cardinalini, Emberizini, Thraupini, Parulini, Icterini), and members of 6 other families of oscines(Turdidae, Monarchidae, Dicaeidae, Sylviidae, Vireonidae, Cor- vidae). The DNA-DNA hybridization data support other evidence indicating that the Hawaiian honeycreepersshared a more recent common ancestorwith the cardue!ine than with any of the other groupsstudied and indicate that this divergenceoccurred in the mid-Miocene, 15-20 million yr ago. The colonizationof the Hawaiian Islandsby the ancestralspecies that radiated to produce the Hawaiian honeycreeperscould have occurredat any time between 20 and 5 million yr ago. Becausethe honeycreeperscaptured so many ecologicalniches, however, it seemslikely that their ancestor was the first passefine to become established in the islands and that it arrived there at the time of, or soon after, its separationfrom the carduelinelineage. If so, this colonist arrived before the present islands from to French Frigate Shoal were formed by the volcanic"hot-spot" now under the island of Hawaii. Therefore,the ancestral drepaninine may have colonizedone or more of the older Hawaiian Islandsand/or Emperor ,which also were formed over the "hot-spot" and which reachedtheir present positions as the result of tectonic crustal movement. The cardueline-drepanininelineage probably diverged from the Fringilla lineage in the late Oligocene, from the New World nine-primaried oscines in the early Oligocene, and from the other oscines in the early Eocene. During this study we also obtained evidence that the vireos (Vireonidae) are not closely related to the New World nine-primaried oscines. Received2 March 1981, accepted1 July 1981.

THE volcanic islands of the Hawaiian chain cific Ocean from Hawaii in the southeast to and the Galapagosarchipelago are well-known Midway and Kure atolls at the northwest end as natural laboratories in which to observe the of the Hawaiian Ridge (see Fig. 3). Hawaii is resultsof the processesof colonization,specia- 3,000 km from California and the Aleutian Is- tion, and . Both island lands, and is 2,500 km from the groupsare substantialdistances from the near- Aleutians and 3,000 km from Kamchatka. The est continent, are composedof many islands, Hawaiian Ridge makes an elbow bend and ex- various distancesapart, and have a variety of tends northward beyond the islands of the ecologicalconditions. The Galapagosarchipel- Hawaiian chain as a line of submergedvolca- ago includes 10 main islands and 6 smaller nos, the Emperor Seamounts. ones located on the equator about 900 km west Twenty-two of Hawaiian honey- of Ecuador. Thirteen species of Galapagos creepers(Drepaninini of Sibley 1970: 99) are finches (,Cactospiza, Certhidea et al.) known, of which six are recently extinct have evolved from the original emberizine (Greenway 1968, Raikow 1977b). Within his- ancestor (Bowman 1961). toric times only eight of the 25 Hawaiian Is- The Hawaiian Islands extend in a broadly lands have supportedpopulations of Hawaiian linear chain for 2,500 km across the central Pa- honeycreepers. l30 The Auk 99: 130-140. January 1982 January1982] HawaiianHoneycreeper Relationships 131

The 11 genera of drepaninines are character- ian honeycreepers,because there are no ana- ized by a remarkable array of bill types, from tomical characters that preclude the relation- the short, -like bills of Psittirostra, Melam- ship and because the carduelines include prosops,and and the slender bills of speciesmore capableof colonizing the Hawai- Viridonia (=), Hirnatione,and Palrneria, ian Islands than do any of the other groups that to the -like structure of Pseudonestorand have been suggested.Bock (1970) also found the long, decurved bills of Hemignathusand that the tongue apparatus of Ciridopsanna re- . This variation caused early taxono- sembles that of the carduelines, not that of the mists to distribute the Hawaiian honeycreep- coerebines. Sibley (1970) also supported the ers among several families, includ- carduelines as the ancestors of the drepani- ing the finches (Fringillidae), flower-peckers nines, based on comparisons of the electro- (Dicaeidae), and honeyeaters (Meliphagidae). phoretic patterns of their -white proteins. Later, it became obvious that the Hawaiian Raikow studied the muscles of the hind limb honeycreepers are closely related to one (1976) and the forelimb (1977a), and reviewed another and that, as in the Galapagosfinches, and extended the morphological evidence of a single ancestralspecies had given rise to the the origin and of the Hawaiian hon- 22 species,which adapted to the many ecolog- eycreepers(1977b). He concluded(1977b: 116) ical niches that were available to the first col- that the drepaninines "arose from a single onists in these oceanic islands (Areadon 1950, founder species," which "was a primitive Sibley 1970, Raikow 1977b). member of the Carduelines." Raikow suggest- The identity of the ancestral has pro- ed that this ancestral cardueline was a typical duced much debate and speculation, but, be- finch similar to the extant Psittirostra and that cause the Hawaiian honeycreepers are nine- the "-feeding habits arosein Hawaii and primaried oscines,their origin must be sought are merely convergentwith similar conditions among the members of that group. Gadow in other families of ." (1891) consideredonly the nectar-feeding taxa In this paper we present the results of com- as possible ancestors and concluded that the parisons among the homologous nucleotide neotropical coerebine honeycreepersare the sequences of the single-copy DNAs of a closestrelatives of the Hawaiian honeycreepers (Hirnatione sanguinea) and that the (Thraupini) are also re- and other taxa of the oscine , using lated to them. Gadow also noted, however, the technique of DNA-DNA hybridization. that the drepaninines and carduelines share The DNA data provide an index to the amount charactersof the horny palate. Lucas (1894) of genetic divergencebetween taxa and to the suggested that the tongues of the Hawaiian time since that divergence began. It is there- honeycreepers could have been derived from fore possible to relate the phylogeny of the that of the New World oriole Icterus or birds to the geologicalhistory of the Hawaiian- that of the paruline genus Dendroica.Sushkin Emperor island chain. (1929) proposed that the cardueline finches and Hawaiian honeycreeperswere most closelyre- METHODS lated, because he found similarities between The DNA hybridization technique takes advan- the bills, skulls, and horny palatesof Psittiros- tage of the complementary structure of the double- tra and those of the carduelines. Amadon stranded DNA molecule. When double-stranded (1950: 232) reviewed the earlier proposals but DNA in solutionis heated to ca. 100øC,the hydrogen concluded that the Coerebini or Thraupini bonds between A-T and G-C base pairs dissociate, were the more probable ancestors. Beecher and the two strands separate. Under proper condi- (1953) found that the jaw musculatureof Psit- tions, the two single strands will reassociateas the tirostrais like that of the carduelines(e.g. Car- solution cools, becausethe complementarybases podacus),and he noted (p. 312) "the striking "recognize" one another. If the temperature is main- similarity of the Hawaiian finches to the car- tained at a high enough level, e.g. 60øC, comple- mentary base pairing will occur only between long dueline finches in all but " but con- homologous sequences of nucleotides. This is be- cluded that the similarities are due to "parallel cause only long sequencesof complementary bases developmentfrom . . . thraupine stock." will have sufficient bonding strength to form stable Bock (1960:477) argued that the carduelines duplexes at that temperature, and only homologous are the most probable ancestorsof the Hawai- sequencespossess the necessarydegree of comple- 132 $t•LEY AtaDA•,QU•ST [Auk, Vol. 99

lO0 such duplexes will cause them to dissociate at a temperature lower than that required to melt 9O conspecificdouble-stranded DNA. Thus, the prop- erty of sequencerecognition exhibited by homolo- gous sequencesand the decreasedthermal stability 80. of imperfectly matched hybrid sequencesform the basis of the DNA-DNA hybridization technique. 70. The degree of nucleotide sequencehomology be- tween the reassociatedsingle strands of any two DNAs can be determined by measuring the per- z centageof hybridization and the thermal stability of the reassociated duplex molecules. Following is a z synopsis of the technique, which is described in more detail by Sibley and Ahlquist 1981). DNAs were obtained from the nuclei of avian erythrocytesand purified by removing the proteins

30. and RNA according to the procedures of Marmur (1961) and Shields and Straus (1975). The purified DNAs were sheared to an averagelength of ca. 500 20. nucleotidesby sonication and sized by electropho- retic comparisonwith DNA fragments of known size produced by digestion of bacteriophageDNA with bacterial restriction endonucleases (Nathans and

o Smith 1975). Single-copy sequenceswere separated

50 55 60 65 70 75 80 85 90 95 from repetitive sequencesat an equivalent Cot of TEMP. DEGREES G. 1,000 (Kohne 1970: 334) at 50øC in 0.48 M sodium Fig. 1. Thermal dissociation curves of DNA-DNA phosphatebuffer and labeled with •25I(Commorford hybrids in which the Apapane (Himationesanguinea) 1971, Prensky 1976). DNA-DNA hybrids were com- was the labeled species. Curve number I is that of posed of one part (=250 ng) •25I-labeled"tracer" the homologous hybrid, x Himatione, DNA and 1,000 parts (=250 /ag) of sheared, whole, number 2 is that of Himatione x Vestiaria coccinea, "driver" DNA at a concentration of 2 mg/ml in 0.48 number 3 that of Himatione x Viridonia (Loxops)vi- M sodium phosphate buffer. The hybrid combina- tions were heated to 100øC for 10 min to dissociate rens, and number 4 that of Himationex FringiIIacoe- Iebs.The curve labeled A is the average + one stan- homoduplexes into single strands, then incubated dard deviation (SD) for the DNA-DNA hybrids for at least 120 h (=Cot 16,000) at 60øCto permit the between Himatione and the 6 cardueline finches single strandsto form hybrid heteroduplexes. (numbers 4-9) in Table 1; the curve labeled B is the The hybrids were placed on hydroxyapatite col- average + one SD for Himatione x the 18 nine-pri- umns immersed in a temperature-controlled water marled oscines (numbers 11-27) in Table 1; and the bath. The temperature was then raised in 2.5øC in- crements from 55øC to 95øC. At each of the 17 incre- curve labeled C is the average + one SD for Hima- tione X the 6 other oscine taxa (numbers 29-34) in ments the single-stranded DNA was eluted in 10 ml Table 1. of 0.12 M sodium phosphatebuffer. This processwas carried out in a custom-built apparatus that accom- modates 25 hydroxyapatite columns in which the temperature is controlled to within +0.1øC. mentarity. Thus, under appropriate conditions of The radioactivity in each eluted samplewas count- temperatureand salt concentration,conspecific dou- ed in a Packard Model 5220 Auto-Gamma Scintilla- ble-stranded DNA may be thermally dissociated, tion Spectrometer,optimized for •2•I. A teletype unit and, becauseof their inherent properties, the single connectedto the gamma counterprinted out the data strands will reassociateonly with their homologous and also punched a paper tape, which is the entry partners. to the computer program. Similarly, if the single-strandedDNAs of two dif- The computer program determined the best fit of ferent speciesare combined under conditions favor- the experimental data to one of four functions: (1) ing reassociation, "hybrid" double-stranded mole- the Normal, (2) the dual-Normal, (3) the "skewed" cules will form between homologous sequences. Normal, or (4) a modified form of the Fermi-Dirac. These sequenceswill contain mismatched bases as The following parameters were then determined a result of the nucleotide sequencedifferences that from the fitted distribution: (1) the modal tempera- have evolved since the two speciesdiverged from a ture of the raw data points, and (2) the T50H,which common ancestor. The lower bonding strength of is the temperature above which less than 50% of the January1982] HawaiianHoneycreeper Relationships 133

sequenceswill be hybridized and below which more than 50% will be hybridized. This is, approximately, the median divergence point. The Ts0H is also the mode of a homologoushybrid and is equal to the mode of any hybrid if all single-copy sequencesin the two speciescould form stableduplexes under the incubation conditions. In each experimental set, the labeled taxon is hy- bridized with itself (=homologoushybrid) and the differences in degrees Centigrade between its pa- rameters and those of the heterologoushybrids are the delta mode and the delta TsoH values. The TsoH (Bonner et al. 1981) is the same as the %0R of 20• Kohne (1970). E

The normalized percentage of hybridization N=I (NPH) is calculated as the percentage of hybridiza- tion of a heterologoushybrid divided by that of the

homologoushybrid x 100. 40

RESULTS AND DISCUSSION iON

50 Figures 1 and 2 and Table 1 present the data from two experimental sets in which the Apa- pane (Himatione sanguinea)was the labeled species. It is apparent that the cardueline Fig. 2. The relationshipsof the Hawaiian hon- finches are the closest relatives of the Hawaiian eycreepersto someother groupsof oscinepassefines honeycreepersand that the Old World genus (Passeriformes:Passeres). The averagesand one SD Fringilla, the New World nine-primaried os- are given for the delta Ts0H values for each group cines, and other oscine groups are progres- relative to Himationesanguinea. Table 1 containsthe data from which this diagram was constructed.Be- sively more distant from them. Figure 2, based causethis diagram is based upon an incompletema- on the T50H values, depicts the relationships trix of DNA hybridization values, it does not indi- indicated by the DNA hybridization data. cate the relationships among the included groups, It is obvious that the thermal dissociation only their averagedistances from the Hawaiian hon- values of the taxa in each group in Table 1 and eycreepers,as represented by H. sanguinea.The Fig. 1 are essentiallyequidistant from the la- groupswere defined by delta Ts0Hvalues and mor- beled Hawaiian honeycreeper, Himationesan- phologicalcharacters. The basis for the time scaleis guinea. The five speciesof carduelinescluster discussed in the text. around an averagedelta T50Hof 4.3 + 0.1 and an average delta mode of 4.1 + 0.1 from Hi- matione.Similarly, the 15 species(17 measure- of the speciesin each group indicatesthat the ments) of New World nine-primaried oscines average rate of nucleotide substitution has have an averagedelta Ts0Hof 7.3 + 0.3 and an been the same, or nearly the same, in the lin- averagedelta mode of 6.7 + 0.4. The remaining eages within each cluster since the time when seven genera of oscines,belonging to at least the most recent common ancestor of all mem- six different families, have an average delta bers of the cluster branched from the lineage Ts0H of 11.2 + 0.2 and an average delta mode that led to the external reference taxon, Himat- of 10.7 + 0.4 from Himatione. ione. If this were not true, we would expect to This pattern of DNA hybridization values see a more scattered distribution of DNA val- indicates that each of these clusters is a mono- ues for the speciesin each group. phyletic taxon relative to Himatione.Each such These data also suggestthat the same aver- cluster is also a "relative rate test" (Sarich and age rate of nucleotide evolution probably oc- Wilson 1967) in which an external reference curs in all lineages. This is indicated by each speciesis used to compare the rates of change of the clusters,especially by the values for the in members of at least two lineages that di- most distant set in Table 1, i.e. numbers 28- verged first from the externalreference species 34, which includes members of at least six fam- and later from one another. The tight clustering ilies of oscines that last shared a common 134 $•BLE¾AND AHLQUIST [Auk, Vol. 99

TABLE1. DNA-DNA hybridization values.a

Labeled Apapane (Himationesanguinea) hybridized with: Delta T50H Delta mode 1. Apapane (Himationesanguinea) 0 0 2. Iiwi (Vestiaria coccinea) 0.8 0.8 3. Amakihi (Viridonia virens) 1.0 1.0 4. (Carpodacuspurpureus) 4.2 4.0 5. (Spinustristis) 4.2 4.1 6. Red (Loxia curvirostra) 4.3 4.1 7. (Loxia curvirostra) 4.4 4.1 8. Pine (Pinicola enucleator) 4.4 3.9 9. (Carduelisspinus) 4.5 4.3 10. Chaffinch (Fringillacoelebs) 6.5 5.9 11. Cinereous Conebill (Conirostrumcinereum) 6.9 6.2 12. Red-winged Blackbird (Agelaiusphoeniceus) 7.0 6.1 13. Blue Honeycreeper(Cyanerpes cyaneus) 7.0 6.4 14. Orchard Oriole (Icterusspurius) 7.1 6.4 15. Yellow-throatedTanager (Iridosornisanalis) 7.2 6.5 16. White-sided Honeycreeper (Diglossaalbilatera) 7.3 6.4 17. Blue Honeycreeper(Cyanerpes cyaneus) 7.4 6.8 18. Palm (Thraupispalmarum) 7.5 6.2 19. Scrub Tanager (Tangara vitriolina) 7.5 6.8 20. Rufous-cappedWarbler (Basileuterusrufifrons) 7.6 6.8 21. Bananaquit ( Coerebafiaveola) 7.6 6.8 22. American Redstart (Setophagaruticilla) 7.6 7.1 23. Lincoln's Sparrow (Melospizalincolnii) 7.7 6.9 24. Common Yellowthroat (Geothlypistrichas) 7.7 7.0 25. Silver-beakedTanager (Ramphoceluscarbo) 7.8 7.0 26. Silver-beakedTanager (Ramphocelus carbo) 7.8 7.2 27. Common Cardinal (Cardinalis cardinalis) 8.1 7.4 28. Common Starling (Sturnusvulgaris) 10.9 10.2 29. Spot-winged Monarch (Monarchaguttula) 11.0 10.3 30. Black Berrypecker(Melanocharis nigra) 11.0 10.9 31. American Robin (Turdus migratorius) 11.2 10.5 32. Garden (Sylvia borin) 11.3 11.0 33. Red-eyed Vireo (Vireo olivaceus) 11.3 11.1 34. Common Crow (Corvusbrachyrhynchos) 11.6 11.0 a Delta T•oH and delta mode values for DNA-DNA hybrids in which the Apapane (Himationesanguinea) was the labeled species.Numbers 1- 3 are Hawaiian honeycreepers(Drepaninini), numbers4-9 are carduelinefinches (Carduelini), number 10 is a fringilline finch (Fringillini), and numbers13.-27 are New World nine-primaried oscines(Emberizinae). Number 28 (Sturnus)is a member of the Muscicapidae:Sturninae, number 29 of the Monarchidae,number 30 of the Dicaeidae, number 33.of the Muscicapidae:Turdinae, number 32 of the Sylviidae, number 33 of the Vireonidae,and number34 of the Corvidae.These allocations follow Sibley (1970) and Sibleyand Ahlquist(3.980).

ancestor with Hirnatione in the early Tertiary, (see Fitch 1976, Wilson et al. 1977, Doolittle 50-55 million yr ago. These six families 1979) have demonstrated that each protein branched from one another at various times (hence each structural gene) evolves at its own during the Tertiary, but today, relative to Hi- rate and that the rates for different proteins rnatione,they differ from one another by a max- vary ca. 600-fold from the slowest histone to imum of 0.7øC T50H. Even these small differ- the fastestimmunoglobulin (Wilson et al. 1977: ences are at least partly due to experimental 610). An averageprotein of 400 amino acids is error. The basis for the above dating is dis- coded for by a gene composedof 1,200 nucleo- cussed below. tides, but the avian haploid genome contains That DNA may evolve at the same average about 1.6 billion nucleotides of which ca. 60- rate of nucleotide substitution in all of the lin- 70% are in the single-copyfraction. The DNA eages of a major group of organisms appears hybridization values are averages acrossthis at first glance to violate logic as well as what entire number; therefore, they reflect the we observe about the variable rates of change amount of nucleotidesequence divergence due in morphologicalcharacters. The conceptthat to the average evolutionary rate of change in the macromolecules(DNA and proteins) evolve about one billion nucleotidesover various pe- in a "clocklike" manner was first proposed by riods of time. Thus, the uniform average rate Zuckerkandl and Pauling (1962), who suggest- of nucleotide substitution in different lineages ed that proteins evolve at "constant" rates of is simply a statistical phenomenon that is the amino acid replacement. Additional studies product of the averaging of large numbers of January1982] HawaiianHoneycreeper Relationships 135 variables operating under the same con- 800,000 mybp and a delta Tm of 2.08øC= ca. 5 straints. Although each nucleotide, in each in- mybp. Angerer et al. (1976)obtained roughly dividual organism,may be evolvingat its own dated values for comparisonsamong three rate, when averagedover such a large number species of sea urchins (Strongylocentrotus), of events the uniform averagerate is the in- which gave a delta TE of 2.5øC= 10 mybp and evitableresult. Sibley and Ahlquist(1981) have a delta TE of 3.5øC= 15 mybp. By using these alsodiscussed this problem. five dated T•0H and TE values to calibrate the Becausethe DNA hybridizationvalues index Hawaiian honeycreeperdata, we estimatethat the averagerate of nucleotidesubstitution and the two lineagesleading to the living cardue- becausethat averagerate is the same in all lin- lines and drepaninines last shared a common eages, the DNA values are proportional to the ancestor15-20 mybp in the mid-Miocene and relative times of divergence of the taxa being that their common ancestordiverged from the compared. If the DNA values could be cali- line leading to Fringilla in the late Oligocene, brated with reference to an external dating ca. 25-30 mybp. The commonancestor of these source, they would also provide an index to lineagesdiverged from the New World nine- absolute time. Ideally, this would be accom- primaried oscine lineage in the early Oligo- plished by datings of divergencetimes cene, 30-35 mybp, and the common ancestor in the phylogeny of one or more groups. But, of all the above groups separatedfrom the line at least for birds, suchdates are either lacking that led to the otheroscine families in the early or unreliable. In our ratite study (Sibley and Eocene,50-55 mybp. These dates are approx- Ahlquist (1981) we used the geologicaldat- imate and preliminary, but they provide the ing of the rift between Africa and South Amer- basis for an examination of the relationship ica to obtain a rough estimate of the time between this possiblehistory of the Hawaiian of the interruption of the genetic contactbe- honeycreepersand the geologicalhistory of the tween the African and South American popu- Hawaiian Islandsand the EmperorSeamounts. lations of the common ancestor of the ostriches There are more than 50 in the and the rheas. This gave a value of ca. 80 mil- Hawaiian chain, but only 25 are currently lion yr before the present (mybp) for a T•0H above sea level, the others being submerged value of 16, or ca. 5 my/1.0 delta T•0H. seamounts.The 10 "high" islandsfrom Hawaii The relationship between the DNA values to and Necker are mountainous; the and time is expectedto be curvilinear,with the "low" islands, from French Frigate Shoalsto ratio decreasingfrom delta 1:5 my at 80 mybp Kure Atoll, are the remnantsof oncehigh vol- to lower ratios for more recent dichotomies. canos that have been eroded to, or near, sea This is because of the effects of back mutations, level. About 1,100 km northwest of Midway, multiple "hits" at the samesite, and other phe- near latitude 23øN, longitude 172øE, the nomena that are proportional to the degree of Hawaiian Ridge makes an abrupt bend to the sequence divergence between the taxa being north and continuesas the Emperor compared. These factors make the DNA hy- chain of some 30 large submerged volcanos, bridization data nonlinear with time and result most of which began as high, mountainousis- in proportional underestimates of the leg lands (Beverley1979). The Emperor Seamounts lengths (internodes) of trees constructedfrom extend northward for 2,500 km to the juncture these data. Sibley and Ahlquist (1981) have of the Kurile and Aleutian trenches between discussedthis problemin greaterdetail. Kamchatka and the western end of the Aleu- A few dated DNA values are available for tian Islands (Fig. 3). more recent divergencetimes, and they reflect The more than 80 huge volcanos in the the curvilinearity described above. They are Hawaiian-Emperor chain are progressively used under the assumptionthat the average older from the southeast to the northwest, be- rate of DNA change in birds, fruit flies, and cause all were formed as the result of relative sea urchins is essentiallythe same, which may motion between the crustal slab of the Pacific or may not be true. plate and an area of vulcanism currently locat- Hunt (1980)compared the singlecopy DNAs ed under the island of Hawaii, where the "hot of three speciesof Hawaiian fruit flies (Dro- spot," or "melting spot," is indicatedby the sophila) and calculated that a median melting active volcanos of and Kilauea. The point depression (delta TE) of 0.5øC = ca. "hot spot" is fixed in position and the Hawai- 136 S•BLEYA•I> AHLQUIST [Auk, Vol. 99

160= 170' 180ø 170' 760= 150 ø

_..:.:..-

50 ø

.40 ø

Gardner Pinnacle

_3Oø Necker

_2Oø

Fig. 3. The Hawaiian Islands and the Emperor Seamounts.Emergent islands are indicated by solid outlines, submerged seamountsby dotted outlines. The active volcanos of Mauna Loa and Kilauea, on the island of Hawaii, are locatedover the "Hot Spot."

Jan-Emperor chain is the track of the average began as an emergent island where Hawaii rate of about 8.5 cm/year linear movement of now is located. As it moved northwestward, the Pacificplate acrossit (Wilson 1963,Dalrym- it was eroded to sea level and became an atoll. ple et al. 1974, Moberly and Larson 1975, R. L. Shallow water foraminifera of late Oligocene Larson pers. comm.). age, ca. 25-30 mybp, mixed with younger The data in Table 2 demonstratethe age pro- planktonic , have been recovered from gression of the Hawaiian-Emperor volcanos a site on K6ko Seamount. These were living and the approximate dates of the rocks in the during a time when coral reef sands were volcanic shields. The dates at which the islands formed in water depths between 30 and 70 m were emergentand vegetatedwould have been (Hottinger 1975). K6ko probably remained somewhat later and their erosion and submer- emergent, or in shallow water as an atoll, for gencelater still. Wave action and other erosive more than 20 million yr, and volcanicactivity forces can truncate a small volcanorelatively may have recurredlong after the Eocene(Mob- quickly, but large structures,such as K6ko Sea- erly and Larson 1975:953). Other Emperor sea- mount, required several million years to be mounts may have had similar histories.When eroded to a flat-topped "." Coral reefs the ancestorsof the Drepaninini and the Car- build up as a sinks, and the resulting duelini diverged, the islands from Hawaii to atoll remains emergent for additional millions FrenchFrigate Shoal were not yet formed. Lay- of years. For example, K6ko Seamount,which san Island was only 200 km from the "hot- was formed in the middle Eocene,ca. 46 mybp, spot," near the present site of Maul, and Mid- January 1982] Hawaiian HoneycreeperRelationships 137

TA•3LE2. Distancesand agesof the Hawaiian-Emperor volcanos.a

Present distance Volcano Island (km) from Kilauea K-Ar age Kilauea Hawaii 0 0-80,000 y Hawaii 100 200,000y Haleakala Maul 180 800,000y West Maui 260 1.3 myb East Moiokai 290 1.5 my Koolau 350 2.6 my Waianae Oahu 400 3.8 my Kauai 550 5.6 my Nihoa Nihoa 700 7.5 my Necker Necker 1,100 10 my French Frigate Shoal French Frigate Shoal 1,200 11 my Laysan 1,800 19 my Midway Midway 2,500 27 my Kanmu Seamount 3,700 37-40 my Yuryaku Seamount 3,800 42 my K•ko Seamount 4,000 46 my Suiko Seamount 5,500 59-64.7 my Meiji Seamount 6,000 70 my a Presentdistances from Kilaueaand potassium-argon(K-Ar) datingsfor someof the Hawaiian IsIandsand EmperorSeamounts. The K-Ar datedsampies came from various positions and depthsin the voIcanicshieIds. Data from CIague and DaIrympIe(1973), DaIrympIe et al. (1974), LanceIotand Larson(1975), Moberly and Larson(1975), DaIrympIe et aI. (1977),and Jarrardand CIague(1977). b my =miIIion years. way was ca. 1,000 km from the "hot-spot," time between 20 and ca. 5 mybp. The ancestor near the present position of . of the drepanininesmust have been the first K6ko Seamount was near the present position passerine to colonize the Hawaiian Islands, of Midway and Meiji Seamount was near however; otherwise, we should expect to find where K6ko is today. Additional older volca- many of their ecologicalniches occupied by nos probably existed to the north of Meiji but speciesof someother group that alsocolonized have been subducted into the Kurile-Aleutian the islands, for example, the Meliphagidae. trench. There is no proof that the Emperor sea- Therefore, it seemslikely that the time of col- mounts were emergent as recently as 15-20 onization either coincided with or was close to mybp, but some may have been atolls that the time of divergence,and we suggestthat it could have provided a landing site for the an- was the colonization event that caused the di- cestralHawaiian honeycreeper.That this early chotomy between the cardueline and drepan- colonist came from the north or northwest is inine lineages. made probableby at least two facts. First, the A colonization date of 15-20 mybp may seem distance, even now, is less from the Aleutians doubtful, becauseit usually has been assumed and Kamchatka to Kure than from California that the ancestral Hawaiian honeycreeper col- to Hawaii. Second, the only records of "acci- onized the islands within the past 5-6 million dental" passerinesreaching Kure and Midway yr, i.e. since the formation of the island of atolls are for Asiatic or Aleutian taxa, namely Kauai. For example, Baldwin (1953: 387) noted the Barn Swallow (Hirundo rusticagutturalis), that "terrain capableof supporting forestsmay Skylark (Alauda arvensis pekinenis), Water Pipit have existed for as long as five million years (Anthus spinoletta japonicus), Red-throated Pip- ß . . thus ample time has apparentlybeen avail- it (A. cervinus), and Snow (Plectro- able for the evolution of terrestrial birds in phenax nivalis townsendi)(Clapp and Wood- Hawaii." Compared with the history of the ward 1968). The prevailing westerly wind Hawaiian drosophilid fruit flies as proposed pattern further increasesthe probability of col- by Beverley(1980), however, the 15-20 mybp onization from Asia. colonizationdate for the honeycreepersis con- Although the DNA data provide an approx- servative. Beverley used the technique of mi- imate date for the cardueline-drepaninine di- cro-complementfixation to compare a larval chotomy, they do not indicate when the an- hemolymph protein among 18 species of cestral Hawaiian honeycreeper colonized the Hawaiian and 40 species of continental dro- islands. This event could have occurred at any sophilids. The rate of evolution of the protein 138 SIBL•¾AND AHLQUIST [Auk, Vol. 99 was calibrated, and the age of the origin of the occurred within the past 5 million yr on the Hawaiian drosophilids was estimated at 42 "high" islands from Kauai to Hawaii. mybp. If this is correct, the ancestral species The Himationex VireoDNA hybrid in Table must have colonized an island that is now a I has a delta T50H of 11.3, which indicates that submerged Emperor Seamount. At 42 mybp the vireos are not closelyrelated to the New Yuryaku Seamount, like Hawaii today, was World nine-primaried oscines.See Sibley and over the "hot spot," and Meiji Seamountwas Ahlquist (in press)for a more detailed study near the presentlocation of Midway. Beverley of this question. (1979) estimated the approximate original heightsof the seamountsby extrapolatingthe CLASSIFICATION undersea slopes of the truncated mountains. The following classification,somewhat mod- For example, the ancient mountain whose rem- ified from Sibley (1970:99), expressesour pres- nant today is Gardner Pinnacle was 2,100 m ent understandingof the relationshipsamong high when it was formed ca. 13 mybp, and the "New World nine-primaried oscines," not Ojin Seamount,in the southernportion of the all of which are confined to the New World. Emperor chain, was at least 1,500 m above sea level ca. 40 mybp. Such islands would have Family Fringillidae supported forests suitable for passerinebirds Subfamily Fringillinae as well as for drosophilid flies. Carson et al. Tribe Fringillini: chaffinches, bram- (1970: 490-491) note that many Hawaiian hon- blings. eycreepersoccur today in the samehabitats as Subfamily Carduelinae do the drosophilids, and some honeycreepers Tribe Carduelini: goldfinches, cross- feed upon the fruit flies. bills, etc. Beverleyproposed that, following the origi- Tribe Drepaninini: Hawaiian honey- nal colonizationca. 40 mybp, the fruit flies be- creepers. gan the diversification that has produced the Subfamily Emberizinae 350 speciesknow today from the Hawaiian Is- Tribe Cardinalini: cardinals, . lands. As the ancient islands eroded away, Tribe Emberizini: buntings, etc. new islands were formed over the "hot spot," Tribe Thraupini: tanagers, including and the flies colonized them when the ecolog- Tersina. ical conditions permitted. Carson (1970) and Tribe Parulini: wood , includ- Carson et al. (1970) have demonstrated that ing Zeledonia. Drosophilaspecies have made at least22 inter- Tribe Icterini: troupials,blackbirds, etc. island colonizationsamong the present main Hawaiian Islands. ACKNOWLEDGMENTS Beverley'sdating of 42 mybp for the colo- For assistancein the laboratorywe thank C. Bar- nization of the Hawaiian-Emperorchain by the kan, M. Pitcher, N. Snow, and F. C. Sibley. The com- Drosophilidaemay prove to be excessive,but puter programwas written by T. F. Smith. For advice any correctionis unlikely to renderimprobable we are indebted to S. M. Beverley,W. J. Bock,T. I. the 15-20 mybp colonizationdate we propose Bonner, R. J. Britten, H. E. Burr, W. M. Fitch, W. for the Hawaiian honeycreepers.Therefore, Hamilton, R. Holmquist, D. E. Kohne, R. L. Larson, from the factsand speculationspresently avail- P. Molnar, M. Nei, R. J. Raikow, and W. F. Thomp- son. For other assistance we thank J. dupont, P. able, we suggestthe following history for the Garayalde,R. Margalef,J. O'Neill, R. Semba,F. Shel- Hawaiian honeycreepers. don, F. C. Sibley, J. Spendelow, N. and E. Wheel- Approximately in the middle Miocene, 15- wright, and D. Wysham. The laboratorywork was 20 mybp, the ancestor of the Hawaiian hon- supported by Yale University and the National Sci- eycreeperscolonized an island in the Emperor- ence Foundation (DEB 77-02594). Hawaiian chain, probably less than 2,000 km from Asia. The birds spreadto other islandsin LITERATURE CITED the chain and speciated as new islands were AMADON, D. 1950. The Hawaiian honeycreepers formed and moved up the chain. Most of the (Aves, Drepaniidae). Bull. Amer. Mus. Nat. adaptiveradiation that producedthe 22 known Hist. 95: 151-262. speciesof Hawaiian honeycreepersprobably ANGERER,R. C., E. H. DAVIDSON, & R. J. BRITTEN. January1982] HawaiianHoneycreeper Relationships 139

1976. Single copy DNA and structuralgene se- FITCH,W. M. 1976. Molecularevolutionary clocks. quence relationships among four sea urchin Pp. 160-178 in Molecular evolution (F. J. Ayala, species. Chromosoma 56: 213-226. Ed.). Sunderland, Massachusetts, Sinauer As- BALDWIN, P. H. 1953. Annual cycle, environment soc. and evolution in the Hawaiian honeycreepers GADOW, H. 1891. Further remarks on the relation- (Aves: Drepaniidae). Univ. Calif. Publ. Zool. 52: ships of the Drepanididae. Pp. 251-257 in Aves 285-398. Hawaiiensis: the birds of the Sandwich Islands BEECriER,W. 1953. A phylogeny of the oscines.Auk (S.B. Wilson and A. H. Evans, Eds.). London, 70: 270-333. R. H. Porter. BEVERLEY,S. M. 1979. Molecular evolution in Dro- GREENWAY,J. C., JR. 1968. Family Drepanididae. sophila.Unpublished Ph.D dissertation.Berke- Check-list of the birds of the world, vol. 14 (R. ley, California, Univ. California. A. Paynter,Jr., Ed.). Cambridge,Massachusetts, ß 1980. A molecularview of the age and origin Mus. Comp. Zool. of the Hawaiian Drosophila.Abstracts, 2nd In- HOTWINGER,L. 1975. Late Oligocene larger forami- tern. Congr. Syst. Evol. Biol. 112.: Vancouver, nifera from Koko Guyot, site 309. Initial Rept. British Columbia, Univ. Brit. Columbia. Deep Sea Drilling Project32: 825-826. Washing- BOCK,W. J. 1960. The palatine processof the pre- ton, D.C., U.S. Govt. Print. Off. maxilla in the Passeres.Bull. Mus. Comp. Zool. HUNT, J. A. 1980. Molecular comparisonof unique 122: 361-488. DNA in Hawaiian Drosophila. Abstracts, 2nd ß 1970. Morphology of the tongue apparatus Intern. Congr. Syst. Evol. Biol.: 113. Vancouver, of Ciridopsanna (Drepanididae). Ibis 114: 61-78. British Columbia, Univ. Brit. Columbia. BONNER, T. I., R. HEINEMANN, • G. J. TODARO. JARRARD,R. D., & D. A. CLAGUE.1977. Implications 1981. A geographicfactor involved in the evo- of PacificIsland and Seamountages for the ori- lution of the singlecopy DNA sequencesof pri- gin of volcanic chains. Rev. GeophysicsSpace mates. Pp. 293-300 in Evolution today. Proc. Physics 15: 57-76. 2nd Intern. Congr. Syst. Evol. Biol. (G. G. E. KOHNE,D. E. 1970. Evolution of higher-organism Scudder and J. L. Reveal, Eds.). Pittsburgh, DNA. Quart. Rev. Biophysics33: 327-375. Pennsylvania,Hunt Inst. Botan. Document. LANCELOT,Y., & R. L. LARSON. 1975. Sedimentary BOWMAN,R. I. 1961. Morphological differentiation and tectonic evolution of the northwestern Pa- and adaptation in the Galapagos finches. Univ. cific. Initial Rept. Deep Sea Drilling Project 32: Calif. Publ. Zool. 58: 1-326. 925-939. Washington, D.C., U.S. Govt. Print. CARSON, H. L. 1970. Chromosome tracers of the or- Off. igin of species. Science 168: 1414-1418. LUCAS,F. A. 1894. Notes on the anatomy and affin- , D. E. HARDY, H. T. SPIETH, & W. S. STONE. ities of the Coerebidae and other American 1970. The evolutionarybiology of the Hawaiian birds. Proc. U.S. Natl. Mus. 17: 299-312. Drosophilidae. Pp. 437-543 in Essaysin evolu- MARMUR,J. 1961. A procedure for the isolation of tion and geneticsin honor of TheodosiusDob- the deoxyribonucleicacid from micro-organ- zhansky (M. K. Hecht and W. C. Steere, Eds.). isms. J. Mol. Biol. 3: 208-218. New York, Appleton-Century-Crofts. CLAGUE,D. A., & G. B. DALRYMPLE.1973. Age of MOBERLY,R., & R. L. LARSON.1975. Mesozoicmag- Koko Seamount, Emperor Seamount chain. netic anomalies,oceanic plateaus, and seamount chains in the northwestern Pacific Ocean. Initial Earth Planet. Sci. Letters 17: 411-415. CLAPP, R. B., & P. W. WOOVWARV. 1968. New rec- Rept. Deep Sea Drilling Project 32: 945-957. ords of birds from the Hawaiian Leeward Is- Washington, D.C., U.S. Govt. Print. Off. lands. Proc. U.S. Natl. Mus. 124: 1-39. NATHANS, D., & H. O. SMITH. 1975. Restriction en- COMMORFORV, S. L. 1971. Iodination of nucleic donucleasesin the analysisand restructuringof acids in vitro. Biochemistry10: 1993-2000. DNA molecules. Ann. Rev. Biochem. 44: 273- 293. DALRYMPLE,G. B., E. A. SILVER, & E. D. JACKSON. 1974. Origin of the Hawaiian Islands. Pp. 24-36 PRENSKY,W. 1976. The radioiodination of RNA and in Guidebook to the Hawaiian PlanetologyCon- DNA to high specific activities. Pp. 121-152 in ference (R. G. Greeley, Ed.). NASA-Ames Re- Methods in cell biology, vol. 13 (D. M. Prescott, search Center, Publ. Ed.). New York, Academic Press. --, D. g. CLAGUE, & M. g. LANPHERE. 1977. RAIKOW,R. J. 1976. Pelvic appendage royology of Revised age for Midway volcano, Hawaiian vol- the Hawaiian honeycreepers (Drepanididae). canic chain. Earth Planet. Sci. Letters 37: 107- Auk 93: 774-792. 116. ß 1977a. Pectoralappendage myology of the DOOLITTLE,R. F. 1979. Protein evolution. Pp. 1-118 Hawaiian honeycreepers (Drepanididae). Auk in The proteins, vol. 4 (H. Neurath and R. L. 94: 331-342. Hill, Eds.), 3rd. ed. New York, Academic Press. 1977b. The origin and evolution of the 140 $•BLE¾ANt> AIaLQU•s•r [Auk, Vol. 99

Hawaiian honeycreepers(Drepanididae). Living lution today. Proc. 2nd Intern. Congr. Syst. 15: 95-117. Evol. Biol. (G. G. E. Scudder and J. L. Reveal, SAR•CIa,V. M., & A. C. W•rsoN. 1967. Immunolog- Eds.). Pittsburgh, Pennsylvania, Hunt Inst. Bo- ical time scale for hominoid evolution. Science tan. Document. 158: 1200-1203. , & --. In press. The relationships of Sia•E•r>s, G. F., & N. A. S•rRAVS. 1975. DNA-DNA the vireos (Vireoninae) as indicated by DNA- hybridization studies of birds. Evolution 29: DNA hybridization. Wilson Bull. 159-166. SUSIaK•N,P. P. 1929. On the systematic position of Smr•¾, C. G. 1970. A comparative study of the egg- the Drepaniidae. Proc. 6th Internatl. Ornithol. white proteins of passerinebirds. Bull. Peabody Congr.: 379-381. Mus. Nat. Hist. 32: 1-131. W•rsoN, A. C., S.S. CAaLSON,& T. J. W•T•. 1977. --, & J.E. Ahlquist. 1980. The relationships Biochemical evolution. Ann. Rev. Blochem. 46: of the "primitive insect eaters" (Aves: Pas- 573-639. seriformes)as indicated by DNA-DNA hybrid- W•rsoN, J. T. 1963. A possible origin of the Hawai- ization. Proc. 17th Intern. Ornithol. Congr. ian Islands. Can. J. Physics41: 863-870. (Berlin, 1978). ZUCKERLANDL, E., & L. PAULING. 1962. Molecular , & -- 1981. The phylogeny and re- disease,evolution, and genic diversity. Pp. 189- lationships of the ratite birds as indicated by 225 in Horizons in Biochemistry (M. Kasha and DNA-DNA hybridization. Pp. 301-335 in Evo- B. Pullman, Eds.). New York, Academic Press.