Considerations Regarding the Evolution of Hawaiian Animals
WILLIAM A. GOSLINEl
INTEREST IN THE BIOTAS of oceanic islands isof long standing. There are several reasons forthe continuing interest. One is the possibilitythat insular evolution may in some respectsrepresent a small-scale model of what has occurred on continents. Another is that , despiteall the work on the subject, the "hows" andthe "whys" of insular evolution remain inadequately answered. Finally, there is the realization that , with the rapid decimation of nativeinsular habitats , it will soon be impossible tostudy many aspects of the subject (Hubbell,1967) .
The general field of evolution in oceanic animals has been reviewed many times, most recently by Miller (1966) and Carlquist (1 965).Zimmerman's summary (1948) for the Hawaiian terrestrial forms is classic. Here, I shalldeal with only certain facets of the subject, andI shall cite only those references from the tremendous literature most pertinent to the matterat hand . This selective method of presentationhas serious faults , but any attempt to be comprehensive would seem only to obscure the threadsof thought that it is the purpose of the paperto present.
Recently, in writing of land plants, Carlquist (1966 :433) has spoken of an "insularsyndrome of interrelated evolutionary phenomena." Insofar as Hawaiian animals are concerned, what is more striking to me is the diversity of evolutionary results. Such variationsoccur not only between groups but within somegroups as well. For example the evolution ofthe Hawaiian drepaniid finches has been verydifferent from that of the sea birds.
This diversity of evolutionary results couldbe exemplified from various animals groups ,most notably insects. However, I shall not dealwith Hawaiian insects at any length, primarilybecause of unfamiliarity with them but also
1 Department of Zoology, Uni versity of Hawaii,Hon olulu, Hawaii 96822 . Manuscript received June13, 1967.
because at the present time they are the subjectof an intensive continuing investigation (Zimmerman , et al., 1948-; Spieth, 1966: 246).Rather, I shall emphasize the evolutionary problems of three Hawaiian animal groups : the inshore fishes, the achatinellid land snails, andthe drepaniid finches. As an introduction to theproblems involved the evolutionary status ofthese three groups in Hawaii is summarizedbriefly.
The Hawaiian inshore fishes (Gosline, 1958;Gosline and Brock, 1960) form part of amarine biota that is essentially similar to, butsomewhat impoverished, as compared with thatof the Central Pacific islands to the south andwest. There are few conspicuous gaps in theHawaiian marine biota, and, of those that dooccur, at least one-the coral genus Acroporawas present in Hawaiian waters in the past(Menard, Allison and Durham, 1962) . There:seems to be a more or less constant infiltrationof non -resident species into the Hawaiianmarine biota today, some of which have become established (Doty, 1961), while somehave not (Brock, 1948). Intentional introductions of purely marine forms into Hawaiianwaters have been mostly unsuccessful. (By contrast, a number of int roductions into areas ofreduced salinity, e.g., Kaneohe Bay on Oahu,have done quite well.) One of the few that hassucceeded, the "Marquesan sardine," has spreadthroughout the waters of the high HawaiianIslands in a matter of a few years (Murphy,1960). Endemism above the species levelamong Hawaiian fishes is dubious. However,about one third of the inshore species are represented by endemic forms. These can usuallybe distinguished from Central Pacific counterparts in 100% of the individuals (for someexceptions, see Gosline, 1955). Aside from afew expected correlations between morphological traits and the relatively cool Hawaiian watertemperatures (see, for example, Strasburg ,1955), the morphological characters by whichthe Hawaiian endemics differ from their Cen-
267
268
tral Pacific counterparts appear to be of a random nature. Within any family of fishes represented in Hawaii, the endemic forms areoften the most abundant.
In striking contrast with the Hawai ianmarine biota, the native terrestrial biota ishighly disharmonic or unbal anced. Greatgroups of animals, e.g., the amphibians, werecompletely unrepresented, whereas others, e.g.,the land snails and drepaniids, proliferatedgreatly. N ot only new Hawaiian species, butalso new genera and famili es evolved. Amongthe achatinellid land snails, the genus Achatinella is restr icted to the island of Oahu, butsome 100 allopatric forms have been described.N o relationship between the peculiarities ofthese forms and the environment they inhabithas .ever been demonstrated. The drepaniidfinches seem to have evolved in quite a differentway. They inhabited all of the major islandsof the Hawaiian chain and some of the smallerislands as well. The most notable differentati onwithin the group is in beak shape, which isassociated with feeding habits (Baldwin,1953). Several different drepaniids were oftensympatric.
One of the main differences between theterrestrial and marine environments in Hawaiiis in the amount of change caused by man.The terrestrial environment has been largelytransformed, in part directly by man via agriculture, etc., but perhaps more by the indirecteffect of animals and plants which man hasintroduced, intentionally or unintentionally.Many of these introductions have now replacedor are replacing the native biota and are directlyor indirectly responsible for the restrict ion orextinction of native forms.
With this brief background, the question ofevolutionary processes will be discussed.
Basic to the evolution ( or lack of it) thatwill occur on any island is the matter of whichorganisms are there and which are not. Toexist, an organism must first arrive, and it thenmust find an environment in which it can survive and reproduce. Both of these aspects depend in part on the isolation of the islandnot isolation in terms of geograph ical-physicalbarriers alone, but in terms of these in relationto the ability of the organism to cross them.
The day when isolation could be considered
PACIFIC SCIENCE, Vol. XXII, April 1968
a causal factor has long since passed. However,that it is a powerful controlling factor is generally recognized. This control acts in two related ways. First , it determines which organismswill get to an island and which will not. Theselectivity of this filtering factor will increasewith increasing isolation and hence will determine in part the extent to which the islandbiota resembles its parental biota. The greaterthe difference between these two biotas, thegreater will be the change in biological selection pressures on any organism arriving on theisland . This point will be discussed later.
Second, any species that establishes itself onan island should , at least for a while, be preserved from contamination by gene flow fromthe parental populati on. If the recent introduction to Hawaii of numerous species (e .g ., theMarquesan sardine , the African snail, thegarden spider, etc.) is any criterion, the initialimmigrants can build up a population of millions of individuals in a few years. Beyond th ispoint contamination from gene flow from afew subsequent immigrants will probably havelittle effect (Gosline, 1958) . There are, however, certain important except ions to th is statement. If, in the process of building up a population from initial immigrants, the populationbecomes debilitated in some way or loses itsability to cope with diseases or parasites whichlater immigrants may bring with them, thensubsequent immigration may matter a greatdeal.
Of factors actually causing insular evolutiononly two will be considered. One is naturalselection, and the other the series of featuresassociated with small population size.
It is generally agreed that differentation proceeds more rapidly in animal populations onsmall islands than on large ones. The questionis : to what extent is this caused by differencesin the selective forces on small islands, and towhat extent to factors associated directly withsmall population sizes. A rather large body ofdata suggests that many of the peculiarities ofsmall-island forms are not directly selected bythe enviroment. Two examples will suffice.Dowdeswell and Ford ( 1953 and elsewhere)have shown that on the larger islands of theScilly group the spotting on the wing of thebutterfly M alliola [urtina remains about as it
Evolution of Haw aiian Animals-c-Gosi.n-rs
is on the adjacent Cornish mainland. On thesmall islands of the group, however, the number of spots on the wing of the females notonly varies from island to island, but increaseson some and decreases on others. Second, Mertens (1934:116) has pointed out that the sameisland may contain a dwarf form of one reptileand a giant form of another. It would be difficult to postulate environmental factors thatwould select animals in these ways.
If small-popul ation forces are to be postulated for such differences, three possibilitiesmust be considered . The first is the random lossof genes which may occur in small populat ionsby genetic drift (Wright, 1931, etc.). Such afactor would presumably be operative in allsmall populations. A special case of geneticdrift is the phen omenon often called foundereffect. Thi s merely expresses the fact that theoriginal immigrants to an island are frequentlyfew in number , and, whether or not they constitute a representative sample, they can bringwith them only a small proportion of the allelespresent in the parental population (see, forexample, Zimmerman, 1948 :122, 123). Thethird possible small-population factor is whatMayr ( 1954) calls internal selection. In largepopulations where each gene often has manyalleles those which work best as heterozygoteswill tend to be selected; on the other hand , insmall populations there will be a larger prop ortion of homozygotes, and alleles which workbest in the homozygotic condition will tend tobe selected. Thus some shift in intern al selection pressure between large and small populations would be expected.
These small-population factors, acting perse, should affect insular immigrants duringthose initial stages when the population is stillsmall ( Fig. 1). But there appears to be noknown instance in which a change at this stagehas been recorded (d. Mayr, 1954). Furthermore, it is a generally accepted dictum that,other things being equal, the older the islandthe greater will be the differentati on in itsbiota; this implies continuous, not just initial,change.
Then how is the differentation that occurs onislands, and more rapidly on small islands, tobe explained? King (1955) conducted selection experiments for DDT resistance on two
269
cultures of Drosophila melanogaster. After adozen or more generati ons some degree ofDDT resistance began to be built up in bothlines. But, as judged from crossbreeding exper iments, the resistance had been built updifferently in the two lines. King (195 5:314)states: "The manner in which a line could respond to selection was to some extent determined by the genetic nature of the sample fromwhich it started , and having started along onecertain road, it kept on. The inevitable sampling error which occurs when a line is takenfrom a larger population is very likely the anlage of the genetic individuality of the line .This is, of course, an example of the principleof genetic drift. . .." The second example isthat reported on by Dobzhansky and Pavlovsky (1957) . In this experiment ten culturesfrom a specially developed laboratory stock ofDrosophila psendo-obscura were started with20 flies each and compared after 17 monthswith ten other cultures that did not begin witha reduced number of individuals . Th ose stockswhich had started with 20 flies showed morevariation than the controls. Again, Dobzhanskyand Pavlovsky conclude ( 1957:316) : "Although the trait studied (g ene arrangement inthe third chromosome) is subject to powerfulselection pressure , the outcome of the selectionin the experimental populations is conditionedby random genetic drift."
One aspect of these experiments by Kingand by Dobzhansky and Pavlovsky may well beof importance for insular evolution. In bothinstances not only were the original samplessmall, but the selection that was exerted uponthem was far different from the selection of thenatural environment from which the flies came.It is as though the samples in the experimentswere subjected to an intense selection pressureat right angles to the pressures to which theancestral "wild" forms had presumably adaptedthemselves. Possibly some of the alleles intensely selected under the laboratory conditionshad been of only peripheral significance to thewild stocks and hence variably represented inthem. Such alleles would be more subject tosampling error among small founder populations drawn from the parental stock than thosepreviously under intense positive selection pressure.
270 PACIFIC SCIENCE, Vol. XXII, April 1962
FIG. 1. The oretical popul ation size (above) andallele variety ( below) plotted against time in a terrestrial animal that successfully immigrates into theHawaiian Islands for the first time. A, D ate ofarr ival; B, time when pop ulation becomes sufficientlylarge that small-popul ation genetic factors will per secause no further loss of alleles; C, initial peak ofabund ance; and D, subsequent equilibrium. For discussion, see text.
Hawaiian endemic fishes cannot, apparently , beattributed to small -popul ation losses alone isthat in many instances the Hawaiian endemicsare not characterized by a simple increase invariability ( as in Dobzhansky and Pavlovsky'sflies) but rather by new and fairly constantcharacters entirely outside the range of variability of the ancestral populations (as inKing's results) . Presumably such charactersmust have arisen through a reintegration and/or evolution from the ancestral genetic systemvia direct or ind irect selection. The same reasoning would seem to apply to at least some ofEisentraut's melanic lizard popul ations.
Judging by personal observation and common knowledge concerning recent successful terrestri al int roductions of animals to the Hawaiianislands (see also Mead, 1961: 180-1 82 ; Tomich,et al., in press) , there is often (presumablyfollowing a longer or shor ter period of smallnumbers) a tremendous initial build-up and"overshoot" in popul ation numb er (Fig. 1) .During my 18 years in Hawaii this has hap-
Insular selection pressures (except, perhaps,for species int roduced by and dependent onman) are similarly at an angle to those exertedon the mainland parental form. Insofar as theisland biota is different from that whence theimmigrants came, it is inevitable that the biological selection pressures on islands will differ.Any immigrant to an island will leave behindat least some of the predators, competitors,diseases, and parasites that the parental mainland stock had to cope with. On the other handthe ini tial immigrants may well have to adaptto new forms of food, cover, etc. (This will beless true only in degree if a species arrives bya series of island hops .)
There is also evidence that selection pressureson small islands are likely to be more radicallydifferent than they are on large islands. Thus,on Manana Island, a small outlier of Oahuwithout domestic cats, the cat flea (Ctenoceph alides felis f elis) has developed an ectoparasitic existence on rabbits (OryctolaguscllllicltlltS) (Tomich , et al., in press). Again,in the Balearic Islands off Spain , Eisentraut(1 949 ) showed that on the smaller outliers thefood of lizards (Lacerta) differed considerablyfrom that on the main islands . As the normalinsect food became more restricted, these lizardsadded the normally avoided ants to their diets,and on very small island s ate even flower petalsand young plant shoots.
Eisentraut believed that this change in diethad a direct metabolic effect resulting in themelanism frequently found in the small-islandpopulations. To me (d. Dowdeswell and Ford ,1953) it seems more likely that the morphological changes so frequently found in smallpoulations are in part the indirect effect ofaltered selection pressures working with timeon the, in part randomly, depleted gene pool ofsmall popul ations. A gradual reintegration ofsuch a gene pool in response to altered selection pressures would likely involve a change inphenotypic characters that are not themselvesselected . Such an interpretation (d. Mayr,1954; D obzhansky and Pavlovsky, 1957) seemsto me to provide the best available explanationfor the often rather heterogeneous differentiating characteristics of insular endemics, e.g.,H awaiian inshore fishes.
The main reason why the peculiarities of
,IIIII
population Isize
A
t I
variety \
of ' .a lleles \
II
I
A8
c
o
time-
Evolution of H awaiian Animals-s-Gosr.ms 271
~.
~.~.:.~.. . .. . .~ ..
FIG. 2. On a background (dotted contours 's representin g Sewall Wright's adaptive peak concept areshown (hatched areass: above, a theoret ical popul ation just before reaching the initi al maximum (C inFig. 1), and below, the same popul ation after anequilibrium size had been reached (D in Fig. 1) .For discussion, see text.
drepaniid finches, with their various beak types,in mind .
Another possibility is that, following theinitial pop ulation explosion, some factor otherthan food supply develops to keep subsequen tnumbers low. This could be disease or parasitism, some change in other environmentalfeatures, or some other factor which wouldlower the reprod uctive rate. That the reproductive rate may be diminished has been stressedby Lack (1954). Lack deals especially withchanges in egg number in birds. But there is
::::- " ,
. .::::::::'-. ' . . . .
. . . . . ...
.... .... ..
...... ....
. ..' ..' -;'. - .....;.-; '. .. .
..
pened with a garden spider (Argiope appema )and the giant Af rican snail (Acha tina flt/ica) ,among the more conspicuous unintentionallyintroduced forms. If it can be assumed that thiscycle happened in the past with our "native"biota, then certain postulates concerning selection pressures would seem to follow.
First , during the period of initia l buildup ofan introduced form , selection pressure must bevery low. (Apparently the other members of thebiota are not initia lly able to cope with or defendthemselves from the new introduction.) H owever, at some point in the buildup, the population becomes excessive, after which it fallsdrastically to a new fluctuating equilibrium wellbelow the previous maximum. The nature of thefactor that sooner or later kills back the initialovershoot is unkn own in any particular instance.There is no reason to believe it is the same in allcases, or that it may not be a combination offactors. W hat is important to the present argument is that after a period of relaxed selectionduring the population buildup a very severeselection pressure of some sort appears . Some ofthe various possibilities are as follows.
First, the animal may eat out the availablefood supply and then die of starvation. Thisapparently happened to the rabbits introducedto Lisianski Island (Bryan, 1942 :192, 193),and almost but, perhaps significantly, not qui tewith the rabbits on Laysan (Warner, 1963 :6,7; cf . also Tomich, et al., in press).
On a larger island with a more varied biotaa second possible situation might occur afterthe immigrant population had overeaten itsoriginal food supply . Assume that an immigrant adapts itself to an insular food supply asclose as possible to that of its parental stock.Assume that , having adapted itself to this insular food source, the immigrant builds up atremendous pop ulation under greatly relaxedselection pressure. At some point it will overshoot its new food supp ly and a severe competition for food will take place. This selectionmay preserve the best adapted individuals ofthe original immigrant type, if enough of thefood supp ly is left. It may also preserve thoseindividuals that have differen tiated fart hest inthe direction of adapting to a new food source(Fig. 2) . This theoretical possibility has beenset up with the evolut ion of the Hawaiian
272
another method, in plants at least, by which thereplacement rate may be held in check. Ratten bury (1962: 354) has said of New Zealandforms:
"Furthermore, the germination of seeds ofmany native species is a matter of extreme dif ficulty, as is evidenced by the sporadic appearance of seedlings which often seem torequi re special conditi ons for their development. Competent nurserymen have experiencedgreat difficulty in germinating native seeds,often resorting to powerful treatments forbreaking the dormancy. In many cases the viable period is very short."
Under conditions of severe interspecific competition, a reduction in the reproduction rate,however accomplished, would seem to be feasible only to the extent that it enables the speciesto raise a greater number of offspring (Lack,1954) . T o drop below that rate would invitereplacement by competit ors (including possible subsequent immigrants of the same species) . If , however, there is very slightinterspecific compet ition, the reproductive ratemight theoretically fall to and somewhat belowthe maximum possible replacement rate with out immediate harm. In my opinion, this iswhat seems to have occurred in many formsamong the native terrestrial Hawaiian biota .
If this is true, a species that experien ced essentially no natural selection from inf ra- and inter specific competition during the initial increasemight again avoid natural selection after anequilibrium had been reached. In the process,however, the species would of course lose itsaggressiveness, especially as compared withsubsequently introduced rapidly breedingforms .
Possible examples of a loss in fertility , apparently before any overpopulation has takenplace, are provided by the Hawaiian hawk (B IIteo solitarillS) and the Hawaiian crow (C oroustropims) , both of which are, and so far asknown always have been, confined to relativelysmall areas of the single island of H awaii. Inview of the general adaptability of relatedmainland forms this geographic restriction ismost difficult to explain. Possibly here we havea "depauperization of biotype" (Hulten, 1937)arising from small populati on effects. Such a"depauperization" of course could and prob-
PACIFIC SCIENCE, Vol. XXII, Apr il 1968
ably often does result in a lowering of reproductive capacity.
Returning finally to the achatinellid snailsof Oahu , it seems obvious from the isolationof many of the colonies of Achatinella (evenbefore their extensive extermination by theintroduced carnivorous snails Englandina andGonaxis; see Krauss, 1964) that they were notspreading. Presumably they can only have beendeveloped from some more "aggressive" ancestral form (and/ or a less dissected topography than exists today) . But given the morerecent isolation in separate colonies there seemsto be no reason why, in the absence of furthergene flow, each colony should not evolve in itsown way as do so many other small populations, particularly land snails. A basically similar provisional hypothesis has been advancedby Carson (1966 :405 ) to explain the formation of Hawaiian species of Drosophilidae.
ACKNOWLEDMENTS
My acquaintance with this subject has beengained over the years through the kind helpof colleagues too numer ous to mention. Forspecific suggestions and comments on the manuscript I wish sincerely to thank Dr. A. H. Banner, Dr. E. A. Kay, and Dr. C. H. Lamoureuxof the Un iversity of Hawaii, and Dr. Y. Kondoof the Bishop Museum, all of whom have beenresidents of Hawaii and interested in theHawaiian biota for much or most of their lives.
REFERENCES
BALDWIN, P. H. 1953. Annual cycle, environment and evolution in the Hawaiian honeycreepers (Aves: Drepaniidae) . Univ. Calif.Publ. Zool. 52 :285-398, pls. 8-11 , 12 figs.
BROCK, V. E. 1948. An additi on to the fishfauna of the H awaiian Islands . Pacific Sci.2:298.
BRYAN, E. H ., JR. 1942. American Polynesiaand the Hawaiian Chain. T ongg PublishingCo., Honolulu. 253 pp., illus.
CARLQUIST, S. 1965. Island Life . The NaturalHistory Press, Garden City, N ew York.viii + 451 pp ., 7 col. pis ., text figs.
- - - 1966. The biota of long-distance dis-
Evolut ion of Hawaiian Animals-i-G ost.rx z
persal. IV. Geneti c systems in the floras ofoceanic islands . Evolution 20 :433-45 5, 2figs.
CARSON, H . 1. 1966. Chromosomal Races ofD rosophila crucigera from the Islands ofOahu and Kauai, State of Hawaii. Univ.Texas Publ. N o. 6615, pp. 405-41 2, 2 figs.
DOBZHANSKY, T. , and O. PAVLOVSKY. 1957.An experimental study of inter action between genetic drift and natural selection.Evolution 11:311-319, 2 figs.
DOTY, M. S. 1961. A canthoph ora, a possibleinvader of the marine flora of Hawaii. Pacific Sci. 15:547- 552.
DOWDESWELL, W. H. , and E. B. FORD. 1953.The influence of isolation on variability inthe butterfly M aniola [urtina 1. Symp. Soc.Exptl. BioI. 7 :254-273, 2 maps.
EISENTRAUT, M. 1949. Die Eidechsen derspanischen Mit telmeer Inseln. Mitt. Zool.Mus. Berlin 26: 1-225, 10 col. pIs., 46 textfigs.
GOSLINE, W . A. 1955. The inshore fish faun aof Johnston Island, a Central Pacific atoll.Pacific Sci. 9 :442-480, 4 figs.
--- 1958. The nature and evolution of theHawaiian inshore fish fauna. Proceedings ofthe Eighth Pacific Science Congress, Vol. 3,pp . 347- 357, 1 fig .
- -- and V. E. BROCK. 1960. Handbook ofHawaiian Fishes. University of Hawaii Press,Honolulu . ix + 372 pp ., 277 figs.
H UBBELL, T . H. 1967. Systematics and the Internati onal Biological Program. SystematicZoology 16:97-99.
H ULTEN, E. 1937. Outline of the History ofArctic and Boreal Biota during the Quaternary Period. Stockholm. 168 pp .
KING, J. C. 1955. Evidence for the integrationof the gene pool from studies of DDT resistance in Drosophila. Cold Spring HarborSymposia on Quantitative Biology. Vol. 20,pp. 311- 317.
KRAUSS, N . 1. H. 1964 . Investigations on biological control of Giant African ( Achatinafttlica) and other land snails. Nautilus 78:21-27.
LACK, D . 1954. The evolution of reproduction
273
rates. In: Huxley, Hardy and Ford, eds.,Evolution as a Process, pp . 143-156. Collier,N.Y.
MAYR, E. 1954. Change of genetic environment and evolution. In : Huxley, Hardy andFord, eds., Evolution as a Process, pp. 157180, 3 figs. Collier , N.Y.
MEAD, A. R. 1961. The Giant Af rican Snail.Un iversity of Chicago Press, Chicago. xvi +257 pp ., 15 figs.
MENARD, H. W., E. C. ALLISON, and J. W.D URHAM. 1962. A drowned Miocene terrace on the Hawaiian Islands. Science 138:896-897.
MERTENS, R. 1934. Die Insel-Reptilien, ihreAusbreitung, Variation und Artbildung.Zoologica (Stuttgart ) 32: 1-209.
MILLER, A. H. 1966. Anim al Evolution onIslands. In: R. I. Bowman, ed., The Galapagos, pp. 10-1 7. Uni versity of CaliforniaPress.
MURPHY, G. I. 1960. Int roduction of theMarquesan sardine, H arengnl« vittata (Cuvier and Valenciennes), to Hawaii an waters .Pacific Sci. 14: 185- 187.
RATTENBURY, J. A. 1962. Cyclic hybridizationas a survival mechanism in the New Zealandforest flora. Evolut ion 16: 348-363, 5 figs.
SPI ETH, H. T. 1966. Courtship Behavior ofEndemic Hawaiian Drosophila. Univ. TexasPubl. N o. 6615, pp . 245-314.
STRASBURG, D. W. 1955. N orth -south differentiation of blenn iid fishes in the Central Pacific. Pacific Sci. 9 :297- 303.
TOMICH, F. Q., C. H. LAMOUREUX, and N.WILSON. In press. Ecological factors onMan ana Island, Hawaii. Pacific Sci.
WARNER, R. E. 1963. Recent history and ecology of the Laysan Duck. Condor 65 :3-23,8 figs.
WRIGHT, S. 1931. Evolution in Mendelianpopulations. Genetics 16 :97-159.
ZIMMERMAN, E. C. 1948. Insects of Hawaii.Vol. 1. Introduction. Un iversity of HawaiiPress, H onolulu. xvii + 206 pp ., 52 figs.
- - - et al. 1948-. Insects of Hawaii. Elevenvolumes published so far. Un iversity ofHawaii Press, Honolulu .