The Mascarene White-Eye Zosterops Borbonica

INTRA-ISLAND VARIATION IN THE MASCARENE WHITE-EYE ZOSTEROPS BORBONICA

FRANK B. GILL

ORNITHOLOGICAL MONOGRAPHS NO. 12 PUBLISHED BY

THE AMERICAN ORNITHOLOGISTS' UNION INTRA-ISLAND VARIATION IN THE MASCARENE WHITE-EYE ZOSTEROPS BORBONICA

FRANK B. GILL

ORNITHOLOGICAL MONOGRAPHS NO. 12

PUBLISHED BY

THE AMERICAN ORNITHOLOGISTS' UNION •,A.L k

Color forms of Zosterops borbonica on Reunion Island (top to bottom): gray morph; highland brown-headed brown morph; lowland brown- headed brown morph; gray-headed brown morph. ORNITHOLOGICAL MONOGRAPHS

This series,published by the American Ornithologists'Union, has been establishedfor major paperstoo long for inclusionin the Union's journal, The Auk. Publicationhas been made possiblethrough the generosityof Mrs. Carll Tucker and the Marcia Brady Tucker Foundation,Inc. Correspondenceconcerning manuscripts for publication in the series shouldbe addressedto the incomingEditor, Dr. John W. Hardy, Moore Laboratoryof Zoology,Occidental College, Los Angeles,California 90041. Copiesof OrnithologicalMonographs may be orderedfrom the Treasurer of the AOU, Burt L. Monroe, Jr., Box 23447, Anchorage,Kentucky 40223. (See price list on insideback cover.)

OrnithologicalMonographs, No. 12, vi + 66 pp. Editor, Robert M. Mengel SpecialEditor for OrnithologicalMonographs No. 12, Marion Anne Jenkinson

Issued May 4, 1973 Price $2.00 prepaid ($1.60 to AOU Members) Library of CongressCatalogue Card Number 73-78162 Printedby the Allen Press,Inc., Lawrence,Kansas 66044 CONTENTS

ACKNOWLEDGMENTS ...... vi

INTRODUCTION ...... 1

REUNION ISLAND ...... 1

FIELD STUDIES ...... •...... 7

BIOLOGY OF ZOSTEROPS BORBONICA ...... 8

COLOR V•U•L•T•ON IN ZOSr•ROeS sORSOWICA ...... 12 Spec•en exm•ation ...... 12 Basis of the color variation ...... 12 Back coloration ...... 14 Head colorationin brown morphs...... 21 Underart coloration ...... 30 MtNS•L V•T•ON ...... 39 S•c•en measurements...... 39 Statistics ...... 40 •alysis ...... 41 DmCUSS•ON ...... 44 V•iafion in cont•ental white-eyes...... 44 T•onomy of Zosterops borbonica ...... 48 Relation of Zosterops borbonica mauritiana to •e Re•ion Isl•d forms ...... 49 Evolutiona• history of Zosterops borbonica ...... 49 Ch•acter variation and adaptation ...... 51 Character divergence • natural populations ...... 52 SUMM•Y ...... 56

L•Ts• •so ...... 59

Arrs•g I ...... 62

ArrsNDg II ...... 65 ACKNOWLEDGMENTS This studywas made possibleprimarily throughthe cooperationof the governmentand citizensof France,Department de la R•union. In particular, I am gratefulto V. Robin, Directeurdes Affairs G•n•rales,for handlingthe many administrativeformalities with maximum dispatch,and to Armand Barau, ConseilerG•n•ral de la R•union, Harry Gruchet, Conservateurdu Museum d'I-IistoireNaturelie de Saint Denis, and ChristianJouanin of the Museum National d'Histoire Naturelie in Paris, who were a sourceof advice, encouragement,and assistance.Several private landownerson Reunion Island, includingMessrs. A. Barau (Bols Rouge), G. Barau (Beaufonds, Ste.Marie), I. Boyer (La Possession),P. Lougnon(La PetiteFrance), J. de Villineuve (Grand Fond), and the Banque de la R6union,permitted me to studyand collectwhite-eyes on their properties.The frequentcooperation of the Servicedes Eaux et For•ts de la R•union, and in particular Messrs. Moulin, Miguet, and Soroquereis also gratefully acknowledged.Messrs. Malick and Laidet of the Service M•t•orologique de la R•nnion placed climatologicaldata from R6unionat my disposal,and M. ThereseanCadet generouslyprovided most of the plant identificationsmentioned in this study. In Ann Arbor, Michigan,the membersof my doctoralcommittee, Morris Foster,Robert W. Storer,Harrison B. Tordoff, and Warren H. Wagner,pro- vided many a stimulatingsuggestion. I am indebtedalso to R. B. Bartels of the Universityof MichiganComputing Center for making availablethose facilitiesand to William Lnnk who paintedthe color illustration. I have bene- fitted from discussionswith J. Alan Feduccia, Theodore H. Fleming, John P. Hubbard, William C. Preston,and Donald R. Tinkle. Acknowledgment and thanksare due especiallyto my wife, Frances,for her many contributions to this study. This studywas supportedby a National ScienceFoundation Predoctoral Fellowshipin 1967 and 1968 and a National ScienceFoundation Grant (GB 3366) to T. H. Hubbellat The Universityof Michiganfor researchin System- atic and EvolutionaryBiology.

vi INTRODUCTION Geographicvariation reflects the geneticdivergence of conspecificpopula- tionsthrough natural selection;it therebyserves as primary evidenceof the evolutionaryprocess. Although every local populationdiffers from adjacent populationswith respectto the frequencyof somecharacteristics, conspicuous phenotypicdifferentiation on a microgeographicscale is evident primarily in terrestrialand relativelysedentary animal groups. Geographicvariation in highly mobile organismssuch as birds, however, tends to involve great distanceso.r else conspicuousisolation. The characteristicuniformity of most localbird populationsseems to reflectgenetic unity that is the combinedresult of gene flow between adjacentdemes and the resistanceof coadaptedgene poolsto fragmentationby disruptiveselection (Mayr, 1963:296, 361). Yet, the precisenessof avian adaptationto climaticgradients (James, 1970) and the surprisingrapidity of populationcharacter divergence in Passerdomesticus (Johnstonand Selander,1964) continueto emphasizethe overridingim- portanceof natural selectionas the primary determinantof patterns of geographicvariation. A birdpopulation on a smalloceanic island is characteristicallyhomogeneous in color and size. Althoughinter-island differentiation is common,intra-island variation is exceptional. Until the discoveryof four apparent races of a white-eye,Zosterops borbonica, on Reunion Island (Storer and Gill, 1966), Jamaicawas the smallestisland (4,540 squaremiles) known to have geographicallyseparated bird populationssufficiently distinct to be considered subspecies.Reunion Island is only about1,000 squaremiles in area. No other islandwith an area of lessthan 10,000 squaremiles is known to have a poly- typic speciesin its avifauna. On the other hand, many larger islandshave differentiatedbird populations,the number of which tends to increasewith island size. But even on islandsas large as Madagascar(240,000 square miles) and Borneo (290,000 squaremiles), only one or two specieshave as many as four races. Thus island size seemsto impose some constraintson geographicvariation. This studyexamines the patternsof phenotypicvariation in Zosteropsborbonica on Reunion Island and considerstheir evolutionary origin and maintenance.

REUNION ISLAND Reunion Island is one of the MascareneIslands, a group of three oceanic islandsof volcanicorigin in the westernIndian Ocean. Reunion is situated approximately500 mileseast of Madagascar(Figure 1) at latitude21ø Off S and longitude55 ø 30' E, is ellipticalin shape,39 mileslong by 29 mileswide, and encompassesabout 1,000 squaremiles. It is both the largestof the three MascareneIslands and the highest,reaching an elevationof 10,068 feet.

1 2 ORNITHOLOGICAL MONOGRAPHS NO. 12

Mauritius -20 $

R•union

-25 $

INDIAN OCEAN

FretroE 1. Location of the Mascarene islands of Reunion and Mauritius in the western Indian Ocean. The third Mascarene island, Rodriguez, is located 250 miles to the east.

Reunion'sfaunal relationships are with Madagascarand the other islandsin the westernIndian Ocean. The islandwas colonizedprimarily from Mada- gascar(Moreau, 1957:399) although.,judging from the similarityof their faunas,irtterchange between Mauriflus and Reunion islands must have occurred quitefrequently. At least8 speciesof birdsbecame extinot shortly after man's arrivalon Reunionin 1663. Today the land bird avifaunaincludes 11 in- digenousand 16 introducedspecies. Discussions of thesespecies are available in Milon (1951), Berlioz (1946), and Watson et al. (1963). TopographicallyReunion Island is an extremelyrugged island, rising steeply 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 3

ALTITUDE PIT 0 N

GRA N D NEIGES

ST. ST. LEUI BENOIT

FIGURE 2. Cross-section of Reunion Island. Original vegetation above St. Leu as follows (from sea level): savanna, evergreen forest, tamarin forest, heath. from the seawith little roomfor co.astalplains (Figures2 and 3). It is com- posedof two overlappingshield volcanos (Upton and Wadsworth,1966:3): Piton de la Fournaise (also called le Volcan; 8,342 feet), an active volcano forming the southeasternthird of the island, and Piton des Neiges (10,068 feet), an extinct volcano in the northwest. The latter, as a result of "extremedevelopment of amphitheatre-headedvalley erosion" (Upton and Wadsworth,1966:5), is dissectedup to depthsof 2,500 metersby deepgorges and threehuge basin-like "cirques." The youngestlavas erupted from Piton des Neigesbetween 100,000 and 350,000 years ago in the late Pleistocene (Chamalaunand McDougall, 1966). The locationof ReunionIsland at the southernedge of the tropicsensures a rather stable,mild climatewhich is dividedinto a warm rainy seasonfrom Novemberto April, and a cool dry seasonfrom May to October. Mid- summerdays averagetwo and one-halfhours longer than winter days and five to six degreescentigrade warmer. Averagemonthly rainfall on the wind- ward side of the islandis about 100 cm higherin the wettestmonth, March, than in the driestmonth, August. The differenceon the island'sdry side is about 15-20 cm. ReunionIsland's rugged relief promotesmarked local variation in temperature and rainfall (Figure 4). The averagerainfall on the west coast,which i.sin a rain shadow,is lessthan 100 cm a year, whereason the easternwind- ward slopesit reaches800 cm a year. Temperaturedecreases about 7øC with each 1,000 metersaltitude, or about2øC per 1,000 feet. Highlandsover 2,000 metersare 14ø to 16ø colderthan the coast,where temperatures average about 22øC in the summer. Rainfall data and temperaturerecords in the form of ten-yearsummaries for 112 and 30 Reunionlocalities, respectively, were made availableby the 4 ORNITHOLOGICAL MONOGRAPHS NO. 12

ST. DENIS

I Pos ?eBr•l•

:hicots ST. PAUL ST.BENOIT Ja • CIRQUE Petite France • DE • SALAZIE

desPiton Neig,• • ForGt ROSE du BGbour

DE For•t Piaine Mourouvin ST. des Cartes TSvelave Blanc LE

Etarig Sal•_•_ les Bains •,• liv. St. ST. LOUIS Etienne

ST. PI ,o s, i? kilometers PHI L I P PE ST. JOSEPH

FIOURE 3. Reunion Island localities.

ServiceM6t6orologique de la R6union. For a few additional localities for whichrecords have beenkept only a few years,I computedaverages from the annual "BulletinsClimatologiques" pubh'shed by the same service. Many of my collectinglocalities were sufficientlyfar from an actualmeteorological sta- tion that I usedrainfall valuesfrom isohyeticalcharts based on the ten-year summariesand calculated temperatures regressed from existingaltitudinal correlations. Two temperatureshave beenused in this studyfor each locality: the mean maximum of the warmest month of the year (either January or February) and the mean minimum of the coldestmonth of the year (usually August). Altitudeswere fixed to the nearest20 meterseither from existing altitude markers or from a 1:50,000 topographicalmap publishedby the Ministbredes Travaux Publicset des Transports(Institut GeographiqueNa- tional) of France. Correspondingto variationin localclimates is a diversityof habitats.The indigenousvegetation is describedby Rivals (1952) and its distributionon the island before the arrival of man is illustratedin Figures 2 and 5. The 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 5

• < •oc < /

• <•00

<400

• '<500

• >soo

FIeURE 4. Distribution of annual rainfall on Reunion Island, Isohyets are in cm. vegetationof certainstudy areas is describedin Gill (1971). Originally there was a tree savannain the arid northwesternlowlands (below 400 meters elevation), whilea tamarin(Acacia heterophylla) forest rich in epiphytesand ferns occurredin protectedlocalities between 1,200 and 2,000 meters elevation, and high altitude heath was presentabove 2,000 meterselevation. Most of the island, however,was coveredby complexevergreen forest with trees about10 metershigh and rarelyexceeding 15 meters. It includedmany species with irregulardistributions, and could be subdividedon the basis of pre- dominantspecies and prevailingclimate. The indigenousvegetation has been almost completelydestroyed below 900 meters elevation on the w•st side of the island and below 400 meters elevationon the eastside; it hasbeen replaced by sugarcane fields, pasturage, andthe exotic plants that typicallysurround human habitations. Scrub vegetation or thicketsof variousmixtures of heath (Philippia), guava (Psidium), 6 ORNITHOLOGICAL MONOGRAPHS NO. 12

FroORE5. Original vegetation of Reunion Island (adapted from Rivals, 1952). Patterns from the west side of island to the center represent the following: A, savanna; B, evergreen forest; C, tamarin forest; D, heath. The clear area arotmd le Volcan representsrecently formed lava which usually lacks vegetation.

Lantana, and Rubushave replacedthe originalforests in many cutover areas. Althoughthe proportionsof the componentspecies may vary greatly, this vegetationtends to be distinctfrom otherhabitats in that it is relativelyopen with no overheadcanopy and is low and densein over-allstructure. Phys- iognomicallyit is closestto thelow ericaceousvegetation of the high altitudes andoften includes some of thesame plant species. The lowlandravines contain lushgrowths of Eugenia]ambos (Myrtaceae) on the eastside of the island and of $chinusterebenthifolius (Anacardiaceae) and Pithecollobiumdulce (Leguminos.ae)on the dry westside. On the westcoast near Etarig Sal• les Bainsa large tract of forestwith Casuarinaequisetifolia (Casuarinac•ae), Pithecollobiumdulce, Melia azedarach(Meliaceae), and $chinusterebenthi- [oliusas the dominantplants, replaces the otherwisebarren cane fields and goatpasturage of the area. Alongthe southeasterncoast from Ste.Rose to St. Josephrelatively undisturbed forest composedprimarily of Mimusopsspp. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 7

(Sapotaceae)descends to about 200 meterselevation. The recent lava flows and young,poor soilsof this area have not encouragedagricultural expansion. The forestsof Grand Brill6 on the southernslopes of le Volcan representa pioneerstage of colonizationof the new lava and are composedprimarily of Casuarinawith a fern (Acrostichum)understory. Much of the indigenousvegetation at higheraltitudes is still intact although one is never far from somehuman disturbanceor exotic species. Between 900 metersand 1,400 meterselevation on the westernslopes of Reunionthe cultivationof geranium(Pelargonium), for its essentialoils, has led to wide- spreaddamage to the forest,which was cleared in plotsfor plantingand which providesfirewood for the distillationprocess. In someareas at moderateeleva- tions, e.g., la Petite France (1,100 meters) and Plaine des Cafres (1,200 meters), the introducedwhite acacia(Leucaena glauca) is now predominant. The wet centralforests of B6bourare relativelyundisturbed although a dense undergrowth of in.troducedFuchsia coccinia is establishedin much of this area. Finally the steep, relatively inaccessiblecliffs throughout the island, includingthe sidesof the otherwiseheavily populated cirque basins, still retain muchof theiroriginal vegetation. The highaltitude heath vegetation is perhaps the leastchanged of Reunion'soriginal habitats, although some, especially near Plainedes Cafres, has been converted into pasturage.

FIELD STUDIES Expeditionswere made to Reunion Island in November, 1964, and from April to December,1967. The 1964 field work was a preliminarysurvey undertakenwith Robert W. Storerthat resultedin our originalcognizance of the problem. Field work in 1967 included intensivesampling throughout the islandand studiesof the white-eye'shabits, through observations of color- bandedindividuals. I alsomaintained captive Z. borbonicaon Reunion Island andat theUniversity of Michigan. I visited 102 localitieson Reunion Island; these are listed by reference numberin AppendixI and are shownon an outlinemap of the island (Figure 6). The localitynames and coordinateswere taken from the topographical map mentionedearlier. The few specimensof Zosteropsborbonica contained in major collections are labelledonly "ReunionIsland" and consequentlyhad little or no value for the presentstudy of variationwithin the island'sconfines. As a result all specimensused in this study were obtained during the two above-mentioned expeditions.The 1964 specimenshave been depositedin the U.S. National Museumcollections and the restin the collectionsof The Universityof Michi- gan Museum of Zoology. In all, 759 specimensfrom 76 different localities were taken. I would remind thosepersons who may feel someconcern about taking specimensof an endemicbird speciesfrom a small oceanicisland not 8 ORNITHOLOGICAL MONOGRAPHS NO. 12

10 1 12 2

57 3

•17 ^ 16 • 15• ?.4 97 72 18 66 7470 , 20 21 19 2599 23 26 8090 82•

56 101 64 63 IO0 42 62 98 95 61 52 94 51 54 41143

0KILOMETERS 5 I0 5••.4,•.91 • 34 86 • FIGURE 6. Localities visited on Reunion Island (see Appendix I). only that Zosteropsborbonica is abundanton Reunion Island, having un- doubtedlybenefited from human disturbancerather than suffered,but also thaXthe speciescontinues to flourishdespite annual predationby local in- habitants. Exceptfor the monthof September,specimens were taken continuallyfrom May throughDecember. Many of thosetaken during the breedingseason from Octoberto Decemberare badly worn and have consequentlybeen excluded from mostanalyses, as were any otherspecimens in similarcondition.

BIOLOGY OF ZOSTEROPS BORBONICA Zosteropsborbonica is an endemicMascarene species found only on the two main islands,Mauritius and Reunion. Althougha typical white-eyein proportionsand behaviorit is somewhataberrant in plumage coloration, having lost both the familiar white eye-ring and the carotenoiddeposition responsiblefor the characteristicgreens and yellowsof Zosteropids.Further- more,the specieshas acquiredwhite "anklets,"a distinctivewhite rump patch formedby the white upper tail coverts,and white axillary featherswhich are 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 9 frequentlyexposed like sunbirdpectoral tufts. Zosteropsborbonica was once includedin the genusMaIacirops with the gray white-eye(Z. modesto)of the SeychellesIslands, but Moreau (1957:400) maintained that the few plumagepeculiarities distinguishing these species did not merit separaterecog- nition at the genericlevel. He consequentlysynonymized Malacirops with Zosterops (see also Moreau, 1967). To date there is no evidenceas to the affinitiesof Zosteropsborbonica. Color variationin ReunionZosterops borbonica was recognizedin various early works,but its naturewas subsequentlythe sourceof muchl•axonomic confusionand speculation.Pollen and van Dam (1868:74), the first authors to recognizethe extentof the color variation,thought it resultedfrom in- dividualbut not sexualdifferences. Shortly thereafter, Hartlaub ( 1877:95) restrictedthe name"borbonica" to brown-backed,gray-headed forms and, on the basisof two specimensin the Newtoncollections now at CambridgeUni- versity,described (pp. 97-98) a secondspecies, Zosterops E. Newtoni, of which the presumedmale was all gray and •thepresumed female rich brown on boththe breast and underparts. Gadow (1884:195-196) did not recognize ZosteropsE. Newtoni on the groundsthat there was a perfect gradationbe- tween"borbonica" and this form in thesmall series of specimenshe examined, but he consideredthe all-grayform to be the male,the brownform to be the female,and the gray-headed, brown-backed form to be theyoung of Zosterops borbonica,corresponding to male Z. E. Newtoni, female Z. E. Newtoni, and Z. borbonicaof Hartlaub, respectively.Gadow also suggestedthat the brown-backedform couldrepresent the winterplumage of the speciesrather than the adult female. Newton and Newton (1888) criticizedthese con- clusionsand supportedHartlaub's earlier recognition of two species. Patternsof geographicdistribution of the various color forms were first describedby Storerand Gill (1966), who delineatedfour subspecies(see frontispiece)from ReunionIsland: Z. b. borbonica,a gray-headed,brown- backedform of theisland's northern lowlands; Z. b. xerophila,a dry western lowland form which was entirely brown above and whitish below; Z. b. aIopekion,a brown-backed,brown-headed form with dark gray and brown underpartsfrom the central forests; and Z. b. edwardnewtoni,an all grayform from the highlandheath zone. Moreau (1967:333) maintainedthe conven- tional pair of subspecies,Z. b. mauritianaon MauritiusIsland and Z. b. borbonica on Reunion Island. Unlike most remainingendemic Reunion birds, which dependon the remnants of indigenousforest, Z. borbonicais common in most habitats, includinggardens, evergreen forests, Casuarina forest, and highlandheath. But despiteits seeminglybroad habitat tolerances, it is primarilya speciesof openand disturbedareas rather than intact forests;its presentdistribution reflectsthe widespreaddestruction of original habitats. In tracts of un- 10 ORNITHOLOGICAL MONOGRAPHS NO. 12 disturbedinterior forestsit is decidedlyscarce except at the edgesof large clearingsand along streambeds. Like mostwhite-eyes, Z. borbonicais a socialspecies usually encountered in partiesof 6 to 20 individuals. Suchflocks are formed throughoutthe year a}thoughthere is a tendencyfor the birdsto breakup into pairs at the heightof the breedingseason. I rarely observedintraspecific aggression; it never seems to be associatedwith territorial or nest site defense. When groupsof feeding Z. borbonicapass close to activenests they are joinedby the nestingbirds, not chased,and nestsare sometimesbuilt within 50 feet of each other in a loose colonial arrangement. Throughoutthe year individual Zosteropsborbonica frequently huddle closelytogether in pairsor groupswhile resting or engagingin mutualpreening. Suchallopreening has been notedin other speciesof white-eyes(Harrison, 1965:195; Eddinger,1967). I have neverwitnessed any aggressionin Z. borbonicathat was the resu}tof proximity. Rather, a white-eyeavoids the huddlingactions of anotherindividual simply by leaving. Only rarely does a singleresting white-eye refuse an opportunityto huddle,and in the Mascarene forestsit is commonplaceto seetwo white-eyessidle along a branchuntil they touch each other. Z. borbonicafeeds primarily on insects,but eats a variety of fruits particularlyduring the winter. Flower feedingis frequent at certain favored flowers. A detaileddiscussion of borbonica'sfeeding ecology is availablein Gin (1971). Zosterops borbonicacoexists on Reunion Island with' Zosterops o,livacea, whichis an aggressive,nonsocial, specialized flower feeder. Zosteropsolivacea feedsprimarily at flowers,catches insects by fly-catching,and rarely eatsfruit. It is usuallyfound in pairs, not flocks,and defendsselected flowering plants againstboth Z. borbonicaand otherZ. olivacea.Zosterops olivacea also is more closelytied to the complexforests of the centralhighlands where there is a varietyof suitableflowers. Zosteropsolivacea seems to have beenthe first white-eyeto colonizethe MascareneIslands and Zosteropsborbonica arrived sometimelater. The differencesbetween the .two speciesand their probableevolution are discussedin detailelsewhere (Gill, 1971). Breeding in Zosteropsborbonica is seasonalas it is in most Mascarene speciesand occursduring the australsummer. In earlyAugust, 1967, I noteda suddenincrease in testicularsize in thesebirds; enlargementcontinued until full size,averaging about 7 x 5 mm, wasreached in earlyOctober. Preceding this morphologicalindication of breedingreadiness, however, were noticeable changesin behavior.Starting in the lastweek of Julyintraspecific chases with repeateddisplacement were regularin feedingwhite-eye flocks. Cer[ain ap- parentlysubordinate individuals often quiveredtheir wingseven when not beingchased, and aggressive individuals often prominently displayed their white 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 11 axillary tufts and rump patch. Pronouncedvocal activityincluding elaborate singingbegan at aboutthe sametime as•the intrasp,ecific chasing. Nestingbegins in late Septemberand Octoberand continues through Decem- ber, perhapsinto J.anuaryand February. In 1967 I noted the first fledged youngon 14 Octoberand found both fledgedyoung and evidenceof nestlings up to the lastday of field work on 21 December.The peak of breedingactivity seemsto occur in October and November; local residentsconfirm this observa- tion. While I have no evidenceof doublebroods the presenceof many nests with eggsin December,1948 (Milon, 1951:141), suggeststhat somepairs may breedmore than oncea year. Althoughthe generaltiming of breedingwas the samethroughout Reunion Island from Septemberthrough December, there were some indicationsof local variability. I found nestswith eggsin the secondweek of Septemberin the wet lowlandsat Beaufonds(Ste. Marie). Yet in the highlandsat le T•velave ( 1,700 m) and Cilaos (900 m) I did not observeactive nest building until the first week of Octoberdespite regular checks. On the dry west coast there was.an abortive nesting attempt at EtangSa16 les Bains in mid-Septemberthat coincidedwith .anabnormal and temporaryburst of growthof the vegetation, but elsewhereon this coast (e.g., St. Paul) nest building was not observed until the end of October. Thus it seemslikely that localpopulations breed in accordwith local seasonality,which may staggerthe onsetof nestingas much as a month between different localities. Cooperativefeeding of the young takes place at least in certain circum- stances.The only Z. borbonicanest I watchedfor an extendedperiod was onein the forestsat Cilaoswith two youngthat were almostready to fledge. At least four adultsbrought food to theseyoung and, of these,three were collectedand proved to be males with ossifiedskulls and enlargedtestes. Followingfledging of the young,family groupsoften join, at leasttemporarily, the feedingflocks in their area. Isolatedfamily groupsoccasionally included an extra male that activelyfed the young. Similar casesof cooperafivefeeding are knownin a varietyof birds,particularly sedentary species in the tropics, and usuallyinvolve closely related adults (siblings) or youngof the previous year (Lack, 1968:80). In Zosteropsborbonica a complete,prebasic molt follows breeding. Of 83 adult borbonicaspecimens taken in December, 13 were molting body feathers. PresumablyJanuary through March, the warmest monthsof the year, are the monthswhen the majority of the populationmolts, but I have no informationfor thisperiod. Of the 143 spedmenshandled in May, 15 were completingtheir molt with tracesof body molt and moltingfeathers in the crownbeing evident. All borbonicacollected or bandedat this seasonwere in fresh,unworn plumage. No molt of anykind wasevident during June and most of July, but an incomplete(pre-alternate) molt, restrictedprimarily to the 12 ORNITHOLOGICAL MONOGRAPHS NO. 12 head, was taking place in 45 of the 103 borbonicabanded in August and September.This molt resultsin no changeof the plumagecoloration. I have no informationon the extent of the postjuvenal(first prebasic) molt which presumablytakes place from Januaryto March at the sametime as the adult prebasic molt. Except for local concentrationsof individualsat abundantfood sources, populationsof Z. borbonicaappear to be sedentarythroughout the year. First therewere no major changesin the distributionof white-eyesduring my field work from April to December,1967. Also, individualscolor-marked during the winter monthsat severallocalities were resightedat thesesame localities during the nestingseason. For example,of the 44 individualsI color-banded at the edgeof the Forat du B•bour on 30 April and 1 May, 4 were observed duringbrief visitsto this localityin Se•ember and December. At Plaine des ChicotsI color-banded19 borbonicain early May, and resighted4 of these in the samehollow during two hoursof observationin early September.Finally 11 of 37 individualsbanded near Cilaos,in late July and early August,were seenagain between 16 Octoberand 20 Decemberin the sametract of forest. These observationsindicate that while Zosteropsborbonica may be quite mobile within a particular area, local populationsare .sedentary. White-eye predatorsare few on Reunion Island. No owls or small bird hawks are known from the island and the large native harder, Circus aeruginosus,shows no aptitude for catching small forest birds, but feeds insteadon introducedmammals and gallinaceousbirds. In the lowlands,nest predationby rats, an introducedsnake (Lycodon aulis), and an introduced lizard (Calotesversicolor) has been reported by local residents.Although the practiceis officiallydiscouraged, large numbersof white-eyesare still caught for food by the nativesof ReunionIsland with bird lime.

COLOR VARIATION IN ZOSTEROPS BORBONICA

SPECIMEN EXAMINATION I evaluatedcolor charactersvisually againsta standardgray background in north sky1/ght on brightto cloudybright gays in the morningin Ann Arbor, Michigan. I evaluatedeach characterindependently of the others and sub- dividedi,t into a seriesof characterstates that includedas many intermediate categoriesas I could consistentlydistinguish (Table 1). I comparedeach specimento a reference specimenwhich representedeach character state. I alsostudied, through microscopic examination, the distributionand intensity of feather pigments.

BAsis OF THE COLOR VARIATION In Zosteropsborbonica, as in certain other birds, for example, Junco (Miller, 1941:179) and Parus (Frank, 1938:164), color variation depends 1973 GILL: VARIATION IN ZOSTEROPS BORBON1C.4 13

TABLE 1 INDEX TO COLOR CHARACTERSIN Zosterops borbonica

Character Character states Reference specimen x

Back color 0 - pure gray 213 784 1 -gray with slight scatteredbrown 213 524 2 - heavy suffusionof brown in gray--nearly 213 469 a 50:50 mixture 3- brown with scattered gray feathers 213 410 4 - entirely brown 213 779 Head color Crown-forecrown 0 - entirely gray 213 500 1 - gray with scattered brown 213 601 2 - 50:50 gray-brown 213 599 3 - brown with gray wash 213 826 4- strong, pure brown 213 343 Nape 0 - pure gray 213 493 1 - gray with scatteredbrown 213 330 2 - 50:50 gray-brown 213 609 3 - brown with gray wash 213 514 4 '- strong, pure brown 213 608 Face 0 - entirely gray 213 577 1- ocular gray and postauricular gray con- 213 569 tinuous and broad with only slight brown 2- well-defined gray above auriculars to eye 213 536 as well as patch of gray on sidesof neck, though separatedfrom ocular gray by a brown collar 3 - diffuse gray around auriculars only 213 874 4- entirely brown, no discernible gray 213 661 Auriculars 0 - pure gray 213 567 1 -slight brown, especially below eye 213 470 2 - 50:50 brown-gray 213 615 3 -predominately brown but with some gray 213 727 4 - entirely brown 213 392 Underpart color Amount of gray 0- white 213 901 1- 213 512 2- 213 352 3- increasing 213 427 4- qntensities 213 309 5- of gray 213 331 6- 213 415 7- 213 384 8- dark gray 213 419 Amount of brown in breast 0- absent 213 496 1- faint brown wash 213 785 2- strong but narrow band 213 411 3- strong wide band 213 782 4- strong and extensive, including parts of 213 341 throat and belly

University of Michigan Museum of Zoology specimen numbers. 14 ORNITHOLOGICAL MONOGRAPHS NO. 12

FIOURE7. Distributionof pigmentsin a brown (A) and a gray (B) feather of Zosterops borbonica.

onthe presence and amount of twotypes of melanir•s:eumelanins, which are blackishgranules, and phaeomelanins,which are reddishor light brown granules.The eumelaninsof Zosteropsborbonica are type B granules(see Moreau,1957:323) depositedin denseconcentrations in the barbulesand to a lesserextent the rachisof eachfeather, but not the barbs (Figure 7). The phaeomelaninscorrespond to Moreau'stype C granulesand are deposited in the barbs and rachis but not the barbules. The basic difference between gray andbrown in the plumageof Zosteropsborbonica lies simplyin the addition of phaeomelaninsto a featherwith eumelanin to producebrown. Severalimportant aspects of the color variation depend on where phaeomelaninsare deposited in the plumage.Other, essentially independent, aspectsof thecolor variation, particularly the colorof the underparts,result fromvarying concentrations of both melanins. There is no variationin color dueto sexor agedifferences once the definitive plumage is attained.

BACK COLORATION Dimorphism.-•I distinguishedfive categoriesof back colorationin my specimensof Zosteropsborbonica from ReunionIsland (Table 1). Of the 343 adultspecimens from localitieswith morethan onephenotype for back color, 301 (87.7 per cent) are eitherpure gray-backed (Category 0) or pure brown-backed(Category 4), whereasonly 42 specimens(12.3 per cent) are intermediate(Figure 8). Of the42 intermediatespecimens, 30 wereclassified in Category1, 5 were put in Category3, and only 7 (2 per cent) were 1973 GILL: VARIATION IN ZOSTEROPSBORBON1CA 15

o 1 2 3 4

BACK COLORATION

FIormE8. Backcoloration of 343 adultsfrom localitiesknown to havemore than oneback color phenotype. Category 0, "pure"gray morphs; Categories 1-3, intermediates;Category 4, "pure"brown morphs (see Table 1). markedlyintermediate, i.e.,in Category2. Thusthis character is strongly dimorphic,with only a slighttendency towards intermediacy, andintermediacy usuallyinvolves scattered traces of brownpigmentation in otherwise gray- backedindividuals. Pure gray-backed individuals tend to have traces of brown elsewherein the plumage particularly onthe flanks, and specimens in inter- mediateback color categories .show corresponding increases in the intensity of brownin theseareas. Unless otherwise stated, when I usethe terms "gray morphs"and "brown morphs" I am referring toindividuals inCategories 0-1 and 3-4, respectively. Geographicdistribution of gray and brown morphs.--The proportion of graymorphs was ascertained for the 83 localitiesfor which6 or morein- dividuals(specimens and sight records combined) were available. The 45 localitiesat whichgray morphs or intermediateswere recorded comprisea large, continuous region in thecenter of theisland (Figure 9); noneare at coastallocalities. At only onelocality (number 42 at 2,300 meters)did I findonly gray morphs; no specimens were taken but all nine birdsobserved there were assignableto categories0 or 1. In contrast,many lowlandlocalities have only brownmorphs. I countedthe numbers of grayand brown morphs at severallocalities during theearly winter and again at thebeginning of breeding season (Table 2). Somevariation between counts, due primarilyto samplingerror, is present 16 ORNITHOLOGICAL MONOGRAPHS NO. 12

o 0 19 9 0 0

0 5O 2

•9 18 0 2125626, "•27200 7 .0 12 0 .0 ½ i0

16 27

9O 47 31 0TM 61 02

0 5 I0

KILOMETERS 0 0

Fioum• 9. The percentage of gray morphs (including Category I) found at each of 83 major sampling localities where data for six or more individuals were available. and to be expectedin thesedata, but the consistencyof the resultsfor each localityis evidenceof seasonalstability of the ratio of the two morphs. An exceptionto this generalobservation was notedat Nez de Boeuf (Plaine desCafres) at 2,060 meterselevation on the edgeof the Ravine de la Rivigre des Reinparts, where an endemic leguminoustree, Sophora nitida, whose flowersseem especially attractive to white-eyes,was abundant. The relative frequencyof gray morphsat this locality increasedfrom 30 per cent to about 70 per cent betweenAugust and early September,1967, as the $ophora flowerswaned (Figure 10). The stabilizationof the morph ratio after early September,as well as my observationsof white-eyesmoving up and down the adjacentcliffs during the day at the height of the Sophoraflowering, indicatethat brownmorphs move up temporarilyfrom the adjacentlowlands to feed at these flowers. Altitude is the mostimportant factor affectingthe distributionand relative abundanceof gray morphs. A regular altitudinalincrease in the percentage of graymorphs occurred along five transectsI madefrom the coastto the high- 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 17 landsin areasof relativelycontinuous, undissected topography (Figures 11, 12, and 13). On four of the five transectsthe relative abundanceof gray morphs increasedcontinuously with altitude above 1,350 meters. On the fifth transect(St. Benoit to la Plaine desPalmistes--open triangles in Figure 13), the relative abundanceof gray morphsincreased continuously above 720 meterselevation; the predominanceof openscrub, which is physiognomically similarto heath,rather than forestat low altitudesmay have beeninfluential becausescrub vegetation was presentat all non-cirquelocalities below 1,300 meterselevation where gray morphs were recorded (Table 3). This is evidence of specifichabitat preferences by gray morphs. With only one exception (locality 52) high gray morph frequencies(over 60 per cent) were not recorded below 1,600 meters elevation (Table 3). Topographyalso affects the proportions of graymorphs. This canbe seen in samplesfrom insidethe deeplyeroded cirques. The data availablefor the Cilaos and Salaziecirques (Figure 14) showhigh gray morph frequencies at relativelylow elevationsand markedfluctuations in the percentageof gray morphsat differentaltitudes. A characteristicfeature of cirquetopography is the separationof high altitudes,where gray morphs are common,from very low altitudes,where they are normally absent, by sheercliffs. This greatreduction of distancesseparating extreme altitudes promotes unpredictable mixtures of

TABLE 2 SEASONAL STABILITY OF TI-IE PROPORTIONS OF BROWN AND GRAY MORPHS AT VARIOUS LOCALITIES ON REUNION ISLAND

Locality Date Number Number Per cent Type of record of gray of brown gray morphs morphs

4 4-6 May 10 7 58.8 Banded birds 9 September 10 10 50.0 Sight 5 7 June 24 5 82.7 Sight 9 September 44 16 73.3 Sight 16 19 April 2 10 16.7 Sight 20 June 2 10 16.7 Specimens 3 November 2 8 20.0 Sight 27 12 May 0 21 0 Sight and specimens 26 November 0 12 0 Sight and specimens 40 28 April-1 May 15 29 34.1 Banded birds 6 September 10 14 41.6 Sight 58 15 June 33 53 38.4 Sight 26-29 July 11 17 39.3 Banded birds 17 October 10 15 40.0 Sight 73 25 May 9 1 90.0 Specimens 5 November ' 12 1 92.3 Sight and specimens 18 ORNITHOLOGICAL MONOGRAPHS NO. 12

•oo,

•,•b.i •o,

o 0>6 70

• 60 On,' U.ILu E_• 50

o,• :2a: 30 o •[ zo

IO I I I '• I , I 7 17 5 23 22 18 AUG AUG SEP OCT NOV DEC

FmURE10. Temporalchange in the ratioof graymorphs (dashed line) and abundance of $ophoraflowers (solid line) at Nez de Boeuf. The data are for 1967, except those for November are from 1964. colorforms, because any movementup or downthe cliffsby white-eyeswould quicklyalter the situationexisting as a resultof altitudinaleffects. Behavioralinteractions o/gray and brownmorphs.---I have analyzedseveral behavioralfeatures to dexerminewhether brown and gray morphs associate non-randomly.On 9 Septemberon the Plalnedes Chicots, when white-eyes were cominginto breedingcondition but prior to actualnesting, I encountered many isolatedforaging pairs (bisexual?)of white-eyesand recordedthe morphsinvolved. At CitaosI recordedthe morphs involved in all intraspecific chasesnoted early in the breedingseason, as well as for establishedpairs of breedingadults, and pairs of white-eyesengaged in huddlingand allopreening. The expectedfrequencies of variouspossible associations were calculatedfrom theobserved percentages of bothmorphs present in .thetwo areas. The observed andexpected frequencies were then tested for significanceby useof a x2 test (Table 4). Boththe associatedpairs of white-eyesat Plainedes Chicots and the matedpairs at Cilaosproved to be randomwith respectto plumagetype. On the other hand huddlingsand chaseswere recordedmore frequentlybe- tweenlike morphsthan was to be expectedif the partnershad been chosen randomlyfrom the total local population. Interpretationof theseresults is clearlydifficult; however a lack of morphdiscrimination in relationto mating, as is suggestedby the randompairing associations at Cilaos,would resultin the maintenanceof polymorphism. Geneticso/back coloration.--JuvenalZosterops borbonica have patternsof 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 19 upperpartcoloration similar to thoseof the adults. Someare entirelygray correspondingto the pure gray morph adultsand have similarvariations in the presenceof brown in other parts of the plumage. Others are brown and equivalentto brownmorph adults in both back and head coloration. Most of my specimensin juvenal plumage were taken with presumed parentsnear Cilao• where both brown and gray morphs are common, the former predominatingaborn 3:2. I avoided flocks that included several famih'esand insteadconcentrated on finding isolatedfamily groups. After establishingthat the adultspresent were feedingthe young, I collectedthe

ß ß ß ß

0 5 I0

KILOMETERS

FIGURE ! 1. Transects made on Reunion Island. Transectsfrom coastalto highland areas: Open stars, St. Paul to Piton des Epinards via le Petite France (Localities 67, 97, 72, 74, 70, 71, 73 in sequence); Closed stars, lgtangsal• les Bains to Piton des Epinards via le T•velave (Localities 51, 94, 95, 61, 98, 62, 63, 64, 73 in sequence); Squares,St. Denis to Plaine des Chicots (Localities 11, 96, 1, 2, 57, 3, 4, 5 in sequence); Closed triangles, St. Benoit to Plaine des Salazes via Takamaka and Forgt du B•bour (Localities 81, 99, 25, 26, 27, 29, 30, 60 in sequence); Open triangles, St. Benoit to Plaine des Cartes via La Plaine des Palmistes (Localities 81, 90, 31, 33, 35, 52, 38, 41 in sequence). Transects in deeply eroded cirques: Open circles, Cirque de Salazie (Localities 14, 15, 16, 17, 20, 22, 18, 19, 60 in sequence); Closed circles, Cirque de Cilaos (Localities 44, 93, 53, 55, 56, 79, 58, 59, 60 in sequence). 20 ORNITHOLOGICAL MONOGRAPHS NO. 12

-•- a. 70 r• 0

60 r• 50 c.) 40

'• 3o 0

•o

o 2 4 6 8 IO 12 14 16 lB 20 22 24 26 ALTITUDE(meters /100) FleURE 12. Altitudinal relationshipsof gray morph frequencieson west slope transects. See Figure 11 for localities and symbols. entiregroup or asmany as possible.In two suchfamily groupsan extra adult male was helpingthe presumedparents feed the young,but each of these "helpers"was the samein plumagecolor as thepresumed male parent and thus did not affectthe colorrelationships in question.I placedthe juvenalspeci- mensin colorcategories equivalent to thoseused for adults(Table 1), except I omittedCategory 3 becauseI doubtthat it is discerniblein this plumage. The mixed broedscollected (Table 5) suggestthat the geneticbasis of back color may involve a simple pair of alleleswith some modifiers. The production of a brownmorph young by grayparents and the consistentproduction of brownyoung by brownparents suggest that gray is dominantover brown. In conclusion,the sympatricoccurrence of gray and brown morphs at numerousupland localities, the formationof mixed pairs, and the production of mixed broods seem to leave little doubt that the discontinuous variation in back color and associatedcharacters of Zosterops borbonica on Reunion Island conformsto a classicalsituation of geneticpolymorphism. This has been defined as "the occurrencetogether in the same locality of two or more discontinuousforms of a speciesin suchproportions •at the rarest of them cannotbe maintainedmerely by recurrentmutation" (Ford, 1964:84). The altitudinaivariation in the proportionsof the two morphspresumably reflects 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 21

n- n e• 7O O

! )- •' •o A/I I /-'?' ß

ß I ?' h•010 A• I :'" ...... -•..J. /'"'""•, • -'• • .....'....-."' '...... IF....•.- ,•_..•.....•&,• -- _ -• ...... • •.•..._ --• -- • •

ALTIT UD E (me•ers•100) •ZGU• •3. A]titud• re•tio•bip• oE •my morph Ereque•cie• on north •nd e•t tmn•ect•. See •i•ure H Eor ]oc•]itie• •nd a differencein the selectiveadvantage of one genotypeover the other under certain environmental conditions.

HEAD COLORATION IN BROWN MORPHS All gray morphshave gray heads. The color of the head of brown morphs variesfrom entirelygray to entirelybrown. I have consideredfour characters for headcoloration: crown,nape, face, and auriculars(Figure 15), each of whichI subdividedinto five characterstates representing increasing amounts of phaeomelaninpigmentation (Table 1). To obtaina singleHead Character Index (HCI) for each specimen,I snmmedthe valuesof the four character states,thus producing 17 indicesranging from the entirelygray-headed form (HCI = 0) to the entirelybrown-headed form (HCI = 16). Over one-halfof the 592 brownmorph specimensin my collectionscored either 0 or 16, and theremaining 45.3 per centwere distributed among the 15 intermediateindices (Figure 16). Brown pigmentappears in the nape before it doeselsewhere on the head, and, as it increasesin in.tensity on the nape,is graduallyadded to the crown. The appearanceof brownin the auricularsis closelycorrelated with its appoar- ancein the crown,but the face is brownonly in thosespecimens with intense 22 ORNITHOLOGICAL MONOGRAPHS NO. 12

TABLE 3 SYNOPSIS OF GRAY MORPHS LOCALITIES

Locality Altitude Vegetation Sample Per cent numbers (meters) size gray morphs2

60 2,460 Heath 10 60 39 2,300 Heath 12 50 42 2,300 Heath 9 100 73 2,200 Heath 28 92 41 2,060 Heath 112 75 59 2,000 Heath 35 51 64 1,980 Heath 32 47 19 1,940 Heath 11 27 5 1,910 Heath 89 78 71 1,780 Tamarin 103 62 4 1,780 Tamarin and Heath 47 54 30 1,720 Forest 20 10 63 1,700 Tamarin 67 31 8 1,700 Forest 11 36 70 1,660 Tamarin 24 62 6 1,650 Forest 24 21 38 1,600 Heath 48 90 23 1,590 Forest 8 12 3 1,580 Forest 63 2-GH 74 1,550 Tamarin 40 25 18 1,530 Forest 20 5 62 1,440 Forest 36 19 72 1,400 Acacia and Forest 19 21 58 1,400 Forest 79 40 24 1,400 Forest 10 20 40 1,400 Forest 63 38 52 1,380 Acacia and Heath 13 61 89 1,360 Forest 26 50 28 1,360 Forest 37 16 22 1,350 Forest 42 14 29 1,350 Forest 23 13 35 1,250 Forest and Scrub 10 50 20-C 1,160 Scrub 20 20 17 1,100 Gardens and Scrub 11 9 61 1,080 Forest and Scrub 41 2 33 1,060 Acacia and Scrub 11 27 49 1,060 Scrub 15 33 1 1,000 Scrub 33 9-GH 56-C 1,000 Casuarina 31 16 79 900 Scmb 18 44 10 820 Scrub 16 19-GH 31 720 Scrub 31 10-GH 16-C 680 Scrub 27 18 25 640 Scrub 15 7-GH 55-C 300 Casuarina 27 33

x C --• Cirque locality. -"GH • Locality at which the brown morphs have gray heads. 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 23

5O ..0

30 z w 2O w h I0 / ø

o 2 4 6 8 I0 12 14 16 18 20 22 24 26 ALTIT UD E (meters/100) FIOUV,E 14. Altitudinal relationships of gray morph frequencies on cirque transects. See Figure 11 for localities and symbols. brown in the crown. This correlationproduces three categoriesof brown morphsthat are visually distinctin the field: those with all gray heads (HCI 0-2), thosewith discerniblebrown on top of the head(nape and crown) but with gray faces (HCI 3-9), and those with brown heads and faces (HCI 10-16). I calculatedthe mean Head CharacterIndex of all adult brown morph specimensfrom each localiXy(Figure 17). Mean indicesat the brown end of the spectrum(12-16) characterizelocalities o.n the west side of the island from theRivi•re desGalets to the Rivi•re St. Etienne,the cirquesof Cilaosand (probably)Mafate and the island'scentral highlands. Indices at the grayend of the spectrum(0-3.4) characterizelocalities below 1,580 meterselevation on the north and 1,380 meterselevation on the east side of the island. A band of intermediateindices (8-11 ) aboutfive kilomeXerswide connectsthe ranges of the two extremetypes on the easternslope in the interior of the island. This zone of contact extends from Plaine d'Affouch.e and Plaine des Chicots across the Cirque de Salazie,Platoau de Belouve,Plaine des Marsouins,and Plateau de Duverneyto la Plainedes Palmistes and probablythe Rivibrede l'Est. On the northwesteoast the regionof contactis reducedto the rocky bed of the Rivi•re des Galets, about 500 metersacross. Here the mean index changes 24 ORNITHOLOGICAL MONOGRAPHS NO. 12

A 8

FIGURE 15. Areas of the head used to determine the Head Character Index: A, crown; B, nape; C, auriculars; D, face. abruptlyfrom 13.6 on the river'ssouth side (locality 68) to 2.9 on the river's north side (locality 87). Low but intermediateindices (5.0-6.2) also characterizelocalities on the island's southern coast between the Rivi•re St. Etienne and Bois Blanc. Parallelingthe abrupt changeon the northwestcoast at the Rivi•re des Galets is a sharpbreak in the meanindex acrossthe Rivi•re St. Etienne, from 14.1 on the northside (locality 44) to 5.4 on the southside (locality 93). Head colorationalso changes abruptly across a coastallava flow just southof Bois Blanc. Observationsjust to the northof locality47 revealedprimarily pure gray-headedbirds, while on the southside of the lav.aflow all white-eyes observedhad sufficientbrown on the napeto be discernedthrough field glasses. This brown-naped form is characteristicof all coastal localities between Rivi•re St. Etienneand Grand BrillO. Frequentexamples of this form also appearin a disjunctpocket at localities98 and 62 abovele T•velave, resulting in lower mean indicesthere than at surroundinglocalities. Throughoutthe islandthe mean index tendsto increasewith altitude. On the westcoast some gray is presentin the face of mostbirds, while, at higher elevationson .the slopesfaces are completelybrown. On the eastern slopes the increasein meanindex reflects the changefrom lowlandgray-headed forms to highlandbrown-headed forms, including some introgression of brown head color characteristics into localities below the zone of contact. Increasedphenotypic variation is evidentin the zoneof contact(Figure 18) 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 25 and reflectsinterbreeding between the brown-headedand gray-headedforms. Neverthelessover 50 per cent of the specimensfrom localitiesin this zone are typical extremeforms (HCI 0-1 or 15-16), indicatingthe operationof isolating mechanismsof some kind. The most obvious environmental factors correlated with the east-west separationof the gray-headedand brown-headedbrown morphs are the differencesin rainfall (seeFigure 4) and the correspondingdifferences in vegetation. The coastalrange of the brown-headedform betweenthe Rivigre des Galetsand the Rivigre St. Etienneis uniformlyarid receivingless than 100 cmof rain a year,whereas the wet southeastcoast receives over 400 cm a year. High rainfallcharacterizes most of thecoast occupied by thegray-headed form, exceptfor the northwesterncoast, but the climatethere clearlyis ameliorated by extensiveirrigation and gardenplantings. An indicationof differencesin dehydrationtolerances of thesecolor forms wasinadvertently obt .ainedwhen I shipped20 live Zosteropsborbonica from ReunionIsland to Michiganby airplane. The trip, which lasted48 hours, testedthe white-eyestolerance of water deprivationbecause the water drinkers in the shippingbox became clogged. Of the 20 birdsin the shipment,6 were gray morphsfrom le T•velave, 6 were gray-headedbrown morphsfrom Beaufonds(Ste. Marie), and8 werebrown-headed brown morphs from several

210

180

150

120

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

HEAD CHARACTER INDEX Fmum• 16. Number of brown morph specimenswith given Head Character Indices (pure gray head: 0; pure brown head----16). 26 ORNITHOLOGICAL MONOGRAPHS NO. 12

.22

3.7 .67 .55 .9 .O5 2.0 2.9 3.0

9.5 .12

15.4 1.5 15. 14-815.7 1.6 15.9 15.3 2.4

16.0 1.9 15.7 16.014,2 1.6 15.7

15.0 1.1i 12.5 13,0 12.1 14.2 13.0 14.215.8 14.7 13.7 14.4. 14,2

o 5 IO 6.2 5.1

FmURE 17. Geographic variation in head color of adult brown morph specimens. Figures indicated are mean Head Character Indices for each locality sampled. Lines delineate distributions of brown-headed brown morphs (11.5-16.0), gray- headed brown morphs (0-3.4), a zone of contact of these two (8.0-10.6), and brown-naped brown morphs (5.0-6.2). locationson the westslopes. Only 2 of eachof the first two categoriessurvived the .trip, whereas 7 of the 8 brown-headedindividu, als were alive, though dehydrated,upon arrival at Detroit, Michigan. The probability that such differencesin survival would occur by chance alone is less than 1 in 10 (Table 6). Brown-headedbrown morphsrange to the highestpart of Reunion Islan.d, while brown morphswith gray headshave a more limited altitudinaldistribution, usuallybelow 1,380 meterselevation. Thus brown-headed brown morphs replacegray-headed brown morphs at high altitudesas well as in the arid lowlands,and seemto occupya wider rangeof climaticextremes. The selective advantageof brown-headedbrown morphsappears to be loweredat the wet middleelevations on the westside of the island,e.g., localities98 and 62 above le T6velave,where I found small numbersof birds that had brown napesonly includedin the localpopulations. 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 27

TABLE 4 BEHAVIORALINTERACTIONS OF GRAY AND BROWNMORPHS OF Zosteropsborbonica

Association Within morph Between morph Gray Brown Total X X Gray Brown

Prebreeding associations Observed 6 2 8 4 (Plaine des Chicots) .0003 Expected 7.3 4.7 Intraspecific Chasing Observed 3 12 15 4 4.19 (Cilaos) 5.514' Expected 10.3 8.7 Mated pairs (Cilaos) Observed 3 8 11 4 1.02 Expected 8.1 6.9 Huddling pairs (Cilaos) Observed 6 17 23 3 29.84*** Expected 14.1 11.9

* P < .05.

*** P < .001.

TABLE 5 BACKCOLOR OF ADULT Zosterops borbonica AND THEIR PRESUMEDOFFSPRING

Back color of Number of Number of offspring of parentsX pairs each back color• d • 0 1 2 4

0 X 1 1 2 1 X 0 1 1 1 4 X 1 1 1 1 4 X 2 1 1 4 X 4 5 8

See Table 1 for definitions of categories.

TABLE 6 DIFFERENTIALSURVIVAL OF Zosterops borbonica COLOR VARIANTSUNDER CONDITIONS OF EXTREME WATER DEPRIVATION

Color variant Lived Died Totals

Gray morphs and gray-headedbrown morphs 4 8 12 Brown-headedbrown morphs 7 1 8 x • = 3.7 P

.05 • 3.84; P < .10 m 2.71. 28 ORNITHOLOGICAL MONOGRAPHS NO. 12

It shouldbe apparentat th/s point that gray morphsoccur primarily in the high altitude r•angeof the brown-headedbrown morphs. I recordedgray morphsat only five localities(numbers 1, 3, 10, 25, 31) in the range of the gray-headedbrown morph, and at theselocalities they occurat low frequencies rangingfrom 1.7 to 19 per centof the local populations.At locality 1 (1,000 meters) I collecteda pair of recentlyfledged siblings, one of which was a gr.ay morph'while the other had a gray head and brown back; this observationis evidencethat gray morphsand gray-headedbrown morphsare compatible colorphases. The replacementof gray-headedwhite-eyes by brown-napedwhite-eyes near Bois Blanc takesplace in a regionof very high rainfall, but doesnot coincidewith any conspicuousclimatic change (see Figure 4). This situation may relate to the fact that the lush forestson the slopesof le Volcan are moderatelywet forestsof low altitudes (Rivals, 1952), which no longer are presentelsewhere on the island. Also this is the only part of Reunion Island subjectedto volcaniceruptions in recenttimes, and lav.aflows may serve to isolate these forms. In summary(see Figure 19), a brownmorph with a gray head is the form of the northern and eastern lowlands of Reunion Island. Its coastal distribution stopsabruptly at theRivi•re desGalets where it is replacedby a brownmorph with a brown head southof the river, and alsoat the recentlava flows of Gr•and Brfil• whereit is replacedby a brownmorph with a brownnape and grayface. Although this brown-naped form occurs in low frequenciesat scattered localitiesin the rangeof the gray-headedbrown morph, the changefrom the predominanceof one to the other acrossthe lava flows at Grand Brill6 is pronounced. The brown-napedform is the only form of Zosteropsborbonica at lowland localities on the south coast of Reunion Island as far west as the Rivi6re St. Etienne,where it is replacedby the brown-headedform on the opposite sideof the river. The brown-headedform thus occupiesthe entire arid west coast of Reunion Island between the Rivi•re St. Etienne and the Rivi•re des Galets. Both riverbedsare only about 500 metersacross and are dry, except for a smallstream, during most of the year. Thus the coastalcontacts of these forms all involveabrupt changesof phenotypesacross conspicuous geographic barriers. The altitudinaltransitions between these same color forms are not as abrupt. The brown-headedform is found at all altitudes on the west side of the island, but is replacedby the gray-headedbrown morph below 1,580 meterson the

FIGURE 18. Head character variability in zone of contact (Localities 6, 16, 21, 24, 27) compared with adjacent areas (Western localities--18, 19, 20, 58, 56, 60, 30, 23, 29, 40; Eastern localities--12, 14, 25, 26, 99, 31, 81, 45, 46, 85). Sample size in parentheses. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 29

6O A. Western localities (,84}

I B. Zone of contact t37} 2O o lO

LU u 0

6O C. Eastern localities

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 HEAD C HARACT E R INDEX 30 ORNITHOLOGICAL MONOGRAPHS NO. 12

REUNION ISLAND o• , •p RlVlERE LOMI• TI"I{S DES GALETS

GR•AND. •RULE

c FmURE 19. Distribution of brown morphs of various head colors. Zones of contact and overlap are indicated by dots within the crosshatching. In the figures of the birds, stipplesindicate gray areas and crosshatchingindicates brown areas. north and 1,380 meters on the east sides of Reunion Island. The transition betweenthe two takes place in a hybrid zone about five kilometerswide. Parentalphenotypes compose at least50 per cent of the hybridizingpopula- tions, suggestingthe operationof someform of partial reproductiveisolation. I have litfie information from the middle elevations on the south side of the islandwhere the brown-naped form mustalso be replacedby the brown-headed form. Usually only brown morphswith brown headsoccur at high altitudes with the gray morph.

UNDERPART COLORATION Variation in the color of the underpartsinvolves both phaeomelaninand eumelanindeposition. Eumelanins are distributedover the entire plumage of the underpartsand vary in concentrationto such an extent that the color of the underpartsranges from nearlypure white to lead gray. In mostspeci- mens, but particularly those with light underparts, the breast tends to be slightlydarker .than either the throator the belly. Phaeomelaninsare confined primarilyto theflanks and breast and thus are more restricted in theunderpart coloration. While all brown morphshave intenselypigmented reddish-brown flanks,the phaeomelanin deposition in the breastplumage varies greatly. Eumelanins.--Toevaluate the eumelanindeposition in the underpartsI comparedall specimenswith a seriesof nineselected specimens representing gradedintensities of melanindeposition from white (CharacterState 0) to lead 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 31

FIGURE 20. Some reference specimensused to representdifferent degreesof gray color in the underparts. From bottom to top--visual indices 0, 2, 4, 6, 9. 32 ORNITHOLOGICAL MONOGRAPHS NO. 12

'• 20 ...... /&" ^...... Length &'"'"'/ .• ...... No. Barbs / .::' ß Breast ßr' 10 / ...... :. o. BBr'a,st 2

Z 6 z -., 0 A...... /x'" /xBelly 1 0 1 2 3 4 5 6 7 8

VISUAL INDEX

FmURE 21. Relation of the extent of eumelanization in feather structure as indicated microscopicallyto the intensity of underpart gray coloration which is grossly visible. gray (CharacterState 8) (Figure 20). Althoughthe total aspectof the under- partswas included in thissingle evaluation, I gaveslightly greater consideration to the belly colorationthan to the breastbecause of the highervariability of the latter and the disruptiveinfluence of phaeomelanindeposition in somespecimens. Followingthe initial sortingI surveyedeach category for homogeneity andreevaluated incongruous specimens. Then I comparedall specimensagain to their standardsand to the standardsone steplighter and one stepdarker. To checkthe repeatabilityof theseevaluations I randomlydrew 107 specimens,or 13.5 per cent of my total sample,and reevaluatedthem on a single morning,completely independently of the originaldetermin•ations; 59 per cent werereassigned to the originalcategory, 41 per cent were within ---1 category and a singlespecimen was two stepsfrom the original. In view of the fine distinctionsbetween categoriesand the numerousvariables affecting such color determinations (see Bowers, 1956), I think this amount of error is acceptab}eand unlikelyto affect significantlythe major resultswhich follow. Differencesin graywhich can be detectedby grossexamination of specimens correspondat the microscopiclevel to differencesin the extentof melanization of the barbules(Figure 21 ). I quantifiedmy microscopicobservations in two ways. First, the numberof barbs with melanin-bearingbarbtries at their base, on oneside of the rachis,was counted from the tip of the featherinwards. The countwas terminatedwhen downy barbtries replaced those of contourfeather 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 33

21 20 19 Y = 1.94 + .0025 X 18

17 o o o

16 o oo

15 o

14 o o ø o 12

II o I0 9 8 o o 7 6 5 OD •0 0 0 o0 %0 o

4 09 0 0 0

2 I % % % %o o 0 oo •o o

0 I 2 3 4 5 6 7 8

UNDERPART GRAY VISUAL INDEX

FIGURE22. Relationshipof altitude to intensityof gray colorationin the underparts in "brown-naped" white-eyes (Head Character Indices 1-9). Specimens from localities98 and 62 are encircledand have not been includedin the regression analysis(see text for explanation). 34 ORNITHOLOGICAL MONOGRAPHS NO. 12

TABLE 7 RELATION OF ENVIRONMENTAL FACTORS TO VARIATION IN UNDERPART GRAY COLORATIONIN Zosterops borbonica

REGRESSION ANALYSIS

Regressionequations Coefficientof F Degrees determination value of freedom

Gray-headed brown morphs (HCI = 0) 2.66 q- .0021'** Alt. 0.386 70.43*** 114 2.04 + .0019'** Alt. + .0014'* Jan. Rain 0.428 41.63'** 111 2.29 + .1305'* Jan. Rain-.005** Aug. Rain 0.19 13.36'* 111 Brown-naped brown morphs (HCI = 1-9) 1.94 q- .0025*** Alt. 0.364 52.64*** 92 1.15 q- .0024*** Alt. + .0016' Jan. Rain 0.384 28.35*** 91 1.77 q- .0087*** Jan. Rain-.015*** Aug. Rain 0.280 17.73'** 91 Brown-headed brown morphs (HCI = 10-16) 1.41 q- .0022*** Alt. 0.497 258.00*** 261 1.38 + .0018'** Alt. + .0021'** Jan. Rain 0.534 154.90'** 270 1.95 q- .0056*** Jan. Rain-.0016 Aug. Rain 0.315 62.11'** 270 Gray morphs 4.87 + .0007*** Alt. 0.067 6.94* 96 4.53 + .0007** Alt. q- .0006 Jan. Rain 0.075 3.88* 95 6.05- .0012 Jan. Rain + .0051' Aug. Rain 0.048 2.38 95

STANDARD PARTIAL REGRESSION COEFFICIENTS

Gray-headed Brown-naped Brown-headed Gray brown morphs brown morphs brown morphs morphs

Altitude 0.556 0.591 0.580 0.254 Rainfall (Jan.) 0.238 0.141 0.232 0.089 Altitude/Rainfall (Jan.) 2.33 4.19 2.50 2.85

* p < .05.

** p < .01.

*** P < .001. structure.Second, I measuredthe length of thebarbale that contained melanin andexpressed this as a percent of thetotal length of thebarbale. This was done by arbitrarilypicking a full lengthposterior barbule near the baseof the nextto outermostbarb of eachfeather. By applyingsuch a procedureto each feather,the variancedue to locationand wear was minimized. Tlfis pro- cedurewas followed for feathersplucked from four or five randomlychosen skinsfrom eachvisual category and for a minimumof sixfeathers from each skin,three from the breast and three from the belly. A significantcorrelation of thesecriteria with thevisual categories is evident(Figure 21 ). Thereis a strongcorrelation between the intensity of grayin the underparts 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 35

X b.I

0 • •0 0 • 0 A ß • 0

4' ß

,x ß A I ß ß

ß 0

ALTITUDE (meters / I00)

FI•U]•E 23. Relationship of altitude and the intensity of gray coloration of underparts of birds from various localities. The mean index of the sample is used for each locality. Open triangles indicate brown-headed brown morph localities; dosed triangles indicate gray-headed brown morph (including brown nape phenotypes) localities; circles indicate gray morph localities. and altitude (Figure 22), coastalp.opulafion•s of Zosteropsborbonica being whitishbelow while thosein the highlandstend to be dark gray. Someof the variationin this relationshipis the resultof averagedifferences between color forms;for example,white-eyes at westernlocalities average somewhat lighter below than at other localities. Linear regressionsshowing the relationship of underpart coloration to altitudeand rainfall were computedseparately for gray morphs,gray-headed brown morphs (HCI 0), brown-napedindividuals (HCI 1-9) and brown- headedindividuals (HCI 10-16). (For discussionof statisticalprocedures, seep. 40). All of theseregressions are significant,but only weakly so in gray morphs (Table 7). The three brown forms have nearly identical regression coefficients(slope) but different intercepts. Brown-napedindividuals have lighter underpartsthan gray-headed,while brown-headedindividuals are lightestof all. The coefficientsof determinationare similar for brown-naped and gray-headedindividuals, but lower than for brown-headedindividuals, in which the regressionaccounts for nearly one-half of the variance. At only three localities(55, 62, 98) could I detect a significantdifference in the underpartcolorations of coexistingcolor variants (head and back color 36 ORNITHOLOGICAL MONOGRAPHS NO. 12

TABLE 8 RELATIONSHIP OF PHAEOMELANIN DEPOSITION IN THE BREAST AND THE HE•iD CHARACTER INDEX

Head Index of brown coloration in breast Character Index 0 1 2 3 4

11 13 • I 0 0 0 12 13 0 0 0 0 13 18 2 0 0 0 14 27 4 0 0 0 15 27 18 8 4 0 16 42 53 50 31 7

Number of specimens. charactersonly). The samplefrom locality 55 (300 meterselevation) in the Cirque de Cilaosincludes gray morphsthat are significantlydarker-bellied than the brown morphsat this locality. At localities98 and 62, which are adjacent,I foundmany brown-naped white-eyes (HCI 1-9) with whiteunder- parts,which were significantlylighter than the rest of the gray-belliedin- dividualsat thesetwo localities.These particular brown-naped specimens also departfrom the altitudinalrelationship of underpartcolor that characterizes otherbrown-naped birds (Figure 23). The additionof rainfall as an independentvariable increases the coefficient of determinationby severalper cent in all of the forms, and the regression coefficientof this variable by itself is significantin all the cases. When re- gressedby themselvesagainst underpart gray intensity,January and August rainfalls account for less of the variance than the above combination and do not constitutea significantregression in graymorphs. Altitude is at leasttwice as importantas Januaryrainfall in all forms, as indicatedby the ratio of the standardizedpartial regressioncoefficients (Table 7). Like other aspectsof the variationin Zosteropsborbonica the basisof the variationin underpartgray colorationappears to be genetic. I kept together in captivityunder identical environmental conditions 11 Zosteropsborbonica including3 light-belliedlowland forms (Index 3) and 8 highlandbirds (In- dices6-8). Despitethe fact that thesebirds have moltedtheir bodyplumage twicewhile in the samecage, the differenceshave beenmaintained, i.e., their underpartcolorations have not convergedas one would expectif they were determinedepigenetically. Phaeomelanins.The amountof phaeomelanindeposition in the breast plumagewas evaluatedin termsof five categoriesof increasingintensity and distribution(Table 1). Correlationsbetween these categories and the Head CharacterIndices in brown morphscan be seenin Table 8. Phaeomelanins are not evidentin the breastcoloration of any brown morph specimenwith 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 37

a7) (48) (34) (38) (61) (51) (42) (18)

1 2 3 4 5 6 7 8

UNDERPART GRAY INDEX FIGURE24. Relationshipof intensity of brownon thebreast and of grayon theunder- partsin brownmorph specimens with Head CharacterIndices of 11 or more. Verticalline, range; horizontal line, sample mean; black rectangle, one standard deviationof the mean;white rectangle, two standarderrors of the mean;sample sizesare given in parentheses.

a HeadCharacter Index of lessthan 11. Faintbrown pigmentation can be dis- cernedin somespecimens with Head Character Indices of 11to 14,but intense phaeomelanindeposition onthe breast occurs only in brownmorph speeimens withindices of 15 and 16, or, in otherwords, in thosewith completelybrown heads.Strong and extensive rufous color in thebre,ast was present in onlya few pure brown-headedspecimens. An increasein the averageamount of brownin the breastwith increased intensityof grayin theunderparts is apparent in all brownmorph specimens with HCI's of 11 or more(Figure 24). The decreasefrom Category7 to Category8 is not statisticallysignificant. Thereis a largecontinuous region on ReunionIsland (Figure 25) in which the meanlocality indices for breastcoloration are lessthan 1.0, indicatinga predominanceof birds with no brownon their breasts although usually a few individualswith sometraces of phaeomelanindeposition were present. This regionincludes the western side of theisland below 1,400 meters and a long narrowband that presumablyextends up the Rivibredes Galets across the cirqueso• Marateand Salazieto the Plateaude Belouve(Trou de Fer), Takamaka,and eventually the Rivibrede l'Est, Pasde Bellecombe,and the 38 ORNITHOLOGICAL MONOGRAPHS NO. 12

0 to.9 1.0to 1.7 2.1to 2.7

FIOURE 25. Geographic distribution of brown-headed morphs with different amounts of brown in their breast plumage. Figures indicated are mean Breast Brown In- dicesfor each locality sampled.

Rivi•re desRemparts. The sectionof this bandfrom the Cirquede Salazie to Takamaka and Rivi•re de l'Est coincideswith the zone of contact described earlier on the basis of Head Character Indices. Pas de Bellecombe, the Rivi•re desRemparts, and Nez de Boeuf,however, are localities at whichmost brownmorphs have brown heads as in the caseof the westernside of the island. The mean index for amount of brown in the breastincreases as one proceeds inland. Onefinds a greaterpercentage of brown-headedwhite-eyes with brown breastsat high.elevations (above 1,400 meters) on thewestern slopes including the Brfil• de St. Paul and Brfil• de St. Leu, the major portionsof the cirques of Mafate and Cilaos,and the higherwestern slopes of the Cirque de Salazie, as well as the ForSt du B•bour, Plaine des Marsouins, Plateau de Duverney, and an isolatedpocket at the head of the ravine of the Rivi•re des Remparts. Finally,in a ratherrestricted region in the verycen•ter of the islandincluding the upperslopes of the ForStdu B•bour, Plainedes Salazes, Coteau Maigre, CoteauKerveguen, and Cilaosforests (the type localityof Z. b. alopekion 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 39

TABLE 9 AVERAGE INDEX OF BROWN IN BREAST COLORATION •'

Locality Altitude Average index Locality Altitude Average index

44 0 0.0 29 1,350 1.1 51 40 0.2 72 1,400 0.0 68 120 0.2 58 1,400 2.1 53 270 0.0 24 1,400 0.9 55 300 0.3 40 1,400 2.3 94 300 0.0 62 1,440 1.0 54 340 0.2 18 1,530 1.5 95 640 0.1 23 1,570 2.3 16 680 0.8 70 1,660 1.4 26 1,010 0.5 8 1,700 1.7 49 1,060 0.6 63 1,700 1.4 61 1,080 0.3 30 1,720 2.3 97 1,080 0.3 71 1,800 1.0 17 1,100 0.2 64 1,980 0.6 27 1,180 1.2 41 2,060 0.9 35 1,250 1.7 73 2,200 1.5 98 1,330 0.5 39 2,300 0.7 22 1,350 0.8 60 2,460 2.7

Based on specimens with Head Character Indices of 11 or more.

Storer and Gill) virtually all brown-headedwhite-eyes have some brown on their breasts,and often this melanindeposition is intenseand extensive.This area is not higherthan other parts of the island which are characterizedby lower indices(Table 9), but it is distinguishedby being one of the wettest highlandareas on the island.

MENSURAL VARIATION

SPECIMEN MEASUREMENTS All specimensof Zosteropsborbonica were measured by me. Wing length (arc) was measuredto the nearest0.5 millimeter;tail lengthwas obtained by insertinga pair of dividersto the baseof the middle pair of rectricesand measuringto the tip of the longestrectrix; bill lengthwas taken as the distance from the tip of the upper mandibleto the nasofrontalhinge, following the procedureoutlined by Moreau (1957: 326); tarsallength was measuredfrom the back .sideof the middle of the tibiotarsa14arsometatarsaljoint to the lower edgeof the lowestundivided scute on the front of the junctionof .themetatarsus with the base of the middle toe (Baldwin et al., 1931). Weights of fresh specimensfrom which the throat cotton was removedwere recordedto the nearest0.1 gram. In order to reducethe character'svariance and to use a linear ratherthan a cubicfunction, I have usedthe cuberoot of body weightin all analyses.Consequently the term "weight"should be understoodto be the cube 40 ORNITHOLOGICAL MONOGRAPHS NO. 12

TABLE 10 CORRELATIONSOF SIZE CHARACTERSOF Zosterops borbonica•

Character pair Brown morphs Gray morphs

HCI 0-7 HCI 10-16

Wing-tail 0.689' * * 0.785 * * * 0.708' * * (190) (235) (79) Wing-tarsus 0.202 *2 0.472'** 0.305'* (194) (254) (91) Wing-bill 0.012 0.288 **• -0.040 (196) (254) (86) Wing-weight 0.174 0.516'** 0.213 (96) (97) (42) Tail-tarsus 0.174 0.372 ***2 0.314 **2 (184) (235) (82) Tail-bill 0.005 0.222 *• -0.160 (187) (234) (77) Tail-weight 0.193 0.549'** -0.060 (96) (90) (40) Bill-tarsus 0.076 0.419'** 0.163 (190) (255) (91) Bill-weight 0.352'** 0.422'** 0.521'** (94) (95) (41) Tarsus-weight 0.310'* 0.477'** 0.312' (94) (94) (44)

• Coefficients are linear product-moment correlation coefficients. Males and females are not considered separately. Sample sizes indicated in parentheses below each coefficient. Significance indicated by asterisks: * P <0.5, ** P <.01, *** P <.001. Correlation not significant independent of weight. root ratherthan the originalmeasure. Since weights increase during the breeding season,particularly in females,only winter specimenstaken from May to Augustwere usedin analysesinvolving this character. There is no seasonal variation in weightduring this period.

STATISTICS

The correlationcoefficients used in this study are linear product-moment coefficientswhich have been computedon an IBM 360 computerusing The Universityof MichiganStatistical Research Laboratory programs BMD:3D and BMD:3R. Multiple regressionanalyses using the least squaresmethod were performedon an IBM 360 computerusing the UCLA program for Multiple Regressionwith CaseCombinations as revisedand adaptedby the StatisticalResearch Laboratory of the University of Michigan (BMD:3R). Chi-squaretests follow the proceduresof Bailey (1959). The Mann Whitney U Test (Siegel, 1956) was usedfor the comparisonof samplemeans. Esti- matesof samplevariances were not calculatedfor samplesof fewer than five 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 41

TABLE 11 REGRESSIONANALYSES OF ALTITUDINAL SIZE VARIATION IN go$tel'o175bol'Dol•ica

Color form Sample Intercept Regression Computed Coefficient of and character size coefficient F value determination (X10 -a)

Brown Morphs HCI = 0-7 Wing length 187 54.52 0.76 12.22'** 0.062 Tail length• 85 4.97 0.27 1.15 0.007 Tarsal length 196 18.21 0.39 19.76'** 0.092 Cube root of 79 -0.02 -0.04 2.28 0.135 body weight Brown Morphs HCI = 10-16 Wing length 230 54.70 1.60 98.04*** 0.300 Tail length• 229 6.90 0.47 15.64'** 0.024 Tarsal length 252 18.23 0.60 84.37*** 0.252 Bill length• 252 8.27 0.12 3.76 0.012 Cube root of 96 2.75 0.106 14.04'** 0.435 body weight Gray Morphs Wing length 77 56.33 0.407 1.35 0.017 Tail length• 77 6.86 0.390 2.86 0.019 Bill length 90 14.07 -0.246 6.33* 0.066 Cube root of 39 9.21 -2.359 1.340 0.169 body weight

With correlated effects of wing length removed. With correlated effects of tarsal length removed. specimens.The .05 level of significanceis requiredfor acceptanceof null hypotheses. Some caution is clearly necessaryin the interpretation of parametric statisticaltreatment of ordinal data like the color rankingsused in this study. The problems inherent in the assumptionof a normal distribution do not affectthe calculationof regressionequaXions but may negatethe significance of thoseregression coefficients which are closeto the minimal level of a• ceptance.

ANALYSIS Superimposedupon the patternsof color variationin Zosteropsborbonica on Reunion Island are differencesin size and proportions. First, in order to test the interdependenceof size characteristics,I analyzed the correlations betweenall pairs of measurementsseparately for the three major color forms (Table 10). Wing length and tail lengthare stronglycorrelated in all color forms. Wing length (and thus tail length) is significantlycorrelated with 42 ORNITHOLOGICAL MONOGRAPHS NO. 12

•o

z i9

i8 r•

3 4 5 6 ? 8 9 I0 I I [2 [3 14 15 16 17 18 19 2021 22 23 24

ALTITUDE (meters / I00)

FIOURE 26. The relationship of tarsal length and altitude for gray morphs. weight only in brown-headedwhite-eyes (HCI 10-16), while bill and tarsal lengthare significantlycorrelated with weightin all color forms. Correlations betweenother pairs of size charactersare affectedby mutual correlationswith weight. However, the partial correlationcoefficients between bill and tarsal lengthsin brown-headedwhite-eyes, and betweenwing and tarsal lengthsin both brown-headedwhite-eyes and gray morphsare significantindependent of weight. Regressionsof size in relation to altitude differ among color forms of Zosteropsborbonica (Table 11). Wing length increasessignificantly with altitudein brownmorphs at a rate of 0.76 to 1.6 millimetersper 1,000 meters of elevationfrom an averagelength of 54.6 millimeters on the coast. Tail lengthincreases significantly with altitude in brown-headedindividuals, even after the correlatedeffects of wing length have been removed. Tarsal length increasessignificantly with altitudein brownmorphs, but the relationshiptends to be curvilinearin gray morphs (Figure 26) and thereforehas not been analyzedfor this form. Bill length increasessignificantly with altitude in brown-headedindividuals, decreases significantly in length in gray morphs (Figure 27), and showsa stronglycurvilinear relationship in brown morphs with grayishheads (Figure 28). Weight increasessignificantly with altitude only in brown-headedindividuals. The specimensfrom sielectedlocalities whose means are not significantly different and which are consistentwith respectto both altitude and color type have been combinedto providesamples large enoughto permit comparison of the different color forms with respectto size (Figures 29, 30, and 31 ). 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 43

15 14.07 -.00024 X

o o

14 o Do o

3 4 5 6 T 8 9 I0 II 12 13 14 15 16 IT 18 19 2021 22 ;>3 24

ALTITUDE (meters / I00)

FIGURE 27. The relationship of bill length and altitude for gray morphs.

Localities 11, 12, and 14 have been combinedinto a singlelowland sample of the gray-headedbrown morph and localities44 and 51 were combinedinto a lowlandsample of the brown-headedbrown morph. A satisfactoryhighland sampleof the gray-headedbrown morph is not available. Brown morphsfrom highlandlocalities 58, 63, and 30 providea singlesample from the upperlimits of abundanceof the brown-headedbrown morph. In most populationsof Zosteropsborbonica the sexesdiffer in both wing length and tail length, males averaging1.0 to 1.5 millimeterslarger •han females. There were no sexualdifferences in tarsallength in any of the populations examined. Sexual differencesin bill length characterizeonly lowland gray-headedbrown morphs, in which males average 0.5 millimeterslonger than females,and highlandgray morphs,in which males average0.9 millimeterslonger than females. Yet, at Cilaos (locality 58), where they are out- numberedby brown morphs,male and femalegray morphsare not different in bill or wing length. There were no sexualdifferences in bill length in any of the brown-headedbrown morphssamples tested. Lowland gray-headedbrown morphshave significantlylonger bills than lowland brown-headedbrown morphs (Figure 30). The altitudinalcorrela- tionsof wing length,bill length,and tarsallength in the brown-headedbrown morphsresult in highlandpopulations averaging significantly larger than lowland populations.Extreme highlandindividuals have bills the same length as lowlandgray-headed brown morphs. Gray morphsat Cilaoshave wings and tarsi equivalentin length to the brown morphs with which they are sympatric,but significantlyshorter bills, the difference correspondingin 44 ORNITHOLOGICAL MONOGRAPHS NO. 12

16.0 o o

15.5 oo o

15.0 o o

14.5

z 14.0 hi

o o 1:5.5 % o

o o o o I5.0

12.0 ø 0 I 2 3 q 5 6 7 8 9 I0 II 12 13 Iq 15 16

ALTITUDE (meters / I00)

FIOURE28. The relationshipof bill length and altitudefor gray-headedbrown morphs (HCI 0-7). magnitudeto thesexual dimorphism found in highlandgray morph populuations andlowland gray-headed brown morphs. There is no significantintra-insular variationin the tail/wing ratio, which averagesvery closeto 0.74 in all populations of all color forms.

DISCUSSION

VARIATION IN CONTINENTAL WHITE-EYES Marked infraspecificvariation in white-eyeson the African continentwas analyzedby Moreau (1957), and I have usedhis observationsas well as my own examinationof museumspecimens in comparingZ. borbonicawith those forms. The differencesin underpartcoloration in Z. borbonicaare easilyas great as the color differencescharacterizing the palest (senegalensis)and darkest (]acksoni,reichenowi) forms of the widespreadAfrican speciesZosterops senegalensis.Color differencesbetween such recognized subspeciesas Zosteropssenegalensis stuhlmanni and Z. s. iacksoni, are less pronounced than betweenextreme variants of Zosteropsborbonica. The point to be emphasizedhere, therefore, is thatcolor variation in Zosteropsborbonica does not simplyinvolve subtle differences but is of a magnitudecomparable to or greaterthan that foundin the continentalspecies. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 45

Differencesin intensityof carotenoidand particularlymelanin deposition, whichare the primarysource of the colorvariation in Africanwhite-eyes, are often correlatedwith climatic trends (Moreau, 1957:404).' In someforms, Z. senegalensisdemeryi, for example,melanin deposition is greaterin regions withhigh rainfall even in the absenceof parallelchanges in altitudeor latitude, althoughmost other darkly pigmentedforms, Z. senegalensisstenocricota, Z. senegalensisstuhlmanni, and Z. senegalensisreichenowi are found at high altitudes.Transitions between these forms and bright yellowlowland populations are evident, but the relative effects of rainfall and altitude have not beenseparated. The presenceof phaeom.elaninshas an irregulardistribution in the Zoster- opidae. Only two African white-eyes,Zosterops pallida and Z. mayo.ttensis, have phaeomelaninpigments in their flank feathers,the usual site of this pigment'sdeposition in zosteropids,although in Z. pallida the extent of phaeomelaninpigmentation in the underpartsis r.eportedto vary greatly (Moreau, 1'957:380). Z. modestafrom the SeychellesIslands appearsto haveminor traces of phaeomelaninsthroughout its plumage,especially in the underparts(pers. obs.). Severalraces of the Japanesewhite-eye, Z. japonica, have rufous-washedsides, and in one closelyrelated Chinesespecies, Z. erythropleura,the flanksare actuallybright reddish-brown. Several races and insularderivatives of Z. lateralis,the widespreadAustralian species, have rustyflanks. The closestparallel to the brownand gray morphs of Zosterops borbonicais found in Zosteropscinerea in Micronesia. On the Caroline Islands,Z. c. cinereais an all graybird very similarto a grayZ. borbonica'in coloration,but on the Palau Islands,Z. c. finschiiis a brownbird resembling the brown morph of Z. borbonica. However, polymorphicplumage characters are virtually unknown and intensebrown plumagesare rare in the Zosteropidae. In Africanwhite-eyes wing lengthincreases significantly in relationboth to increasingaltitude (0.69 mm/1,000 feet of elevation)and decreasing temperature (2.8 mm/10 ø F) (Moreau, 1957:330). In those birds the relationof tail lengthto altitudeseems to be a functionof its high correlation with wing length(correlation coefficient = 0.93) but the tail/wing ratio also increaseswith altitude,i.e., the averagetail lengthbecomes proportionately longerrelative to increasingaverage wing length. Bill lengthincreases slightly with decreasingminimum temperature, even after highly correl.atedeffects of wing length (correlationcoefficient = 0.83) are removed. Thus white- eyesinhabiting higher or coolerparts of Africa tend to be generallylarger than those in the lowlands. No information is available for variation in tarsal lengthor weightin African white-eyes. The altitudinalrelationship of wing length in Zosteropsborbonica is virtuallythe sameas in the African speciesand it is probablethat similar 46 ORNITHOLOGICAL MONOGRAPHS NO. 12

c I I D

51 52 5• 54 55 56 57 58 59 60 6l

WING LENGTH (rnm)

FIOURE29. Wing length of representativepopulations of Zosterops borbonica on Reunion Island. A. Localities 11, 12, and 14 combined: (1) 15 males, (2) 9 females; B. Localities 44 and 51 combined: (1) 13 males, (2) 10 females; C. Locality 58: (1) 14 brown morph males, (2) ! 1 brown morph females, (3) 12 gray morph males and females; D. Localities 63 and 30 combined (browns only): (1) 13 males, (2) 12 females; E. Localities 41 and 73 combined (grays only): (1) 11 males, (2) 9 females. (Horizontal line, range; vertical line, samplemean; black rectangles, one standard deviation of the mean; open rectangles, two standard errors of the mean.) selectiveagents are involved. On the other hand, there is no altitudinal correlation of the tail/wing ratio in Zosteropsborbonica. Indicationsof such a rela- tionshipin Africa (Moreau, 1957:336) are basedon ratios of averagetail lengthsand average wing lengths rather than on averagesof the ratioscalculated separatelyfor each specimenin the population. The two kinds of ratios are not comparable. Distancesbetween phenotypically distinct populationsof white-eyesin Africa are usuallyon the order of hundredsof miles, but in certain cases,as, 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 47 for example,between Zosterops senegalensis demeryi and Z. s. senegalensis, it appearsthat the belt of transitionis very narrow, perhapsonly a few miles wide (Moreau, 1957:353, 357), although this is poorly documented. Similarly in some areasthe altitudinalclines in size are very steep, as, for example,on the edgeof the CongoBasin, where populations only 20 and 40 miles apart differ markedly in their measurements(Moreau, 1957:368). Severalof the distinctivemontane populations in Kenya are separatedby geographicaland ecologicalgaps of no more than 25 miles (Moreau, 1957: 363). In summary, certain aspectsof the variation in Zosterops borbonica, particularlythe ecogeographicclines in size and intensityof melanindep.osi- tions, parallel well known patternsof infraspecificvariation in continental white-eyes. Other important aspectsof borbonica'scolor variation which relateto the presenceor absenceof phaeomelaninpigments are unusual,if not uniquein the Zosteropidae.Certainly the combinationof suchdiscordant, but striking,patterns of geographicvariation is not known for any other species

A _rim

i) C 2)

12.5 1.3.0 135 14.0 14.5 15.0 15.5

BILL LENGTH (turn)

Fmtat• 30. Bill lengths of representative populations of Zosterops borbonica on Reunion Island. A. Localities I1, 12, 14, and 81 combined: (I) 16 males, (2) I1 females; B. Localities 44 and 51 combined: 28 males and females; C. Locality 58: (I) 26 brown morphs, (2) 13 gray morphs; D. Localities 63 and 30 combined (brown morphs only): 23 males and females; E. Localities 41 and 73 combined (gray morphs only): (I) 17 males, (2) I0 females. 48 ORNITHOLOGICAL MONOGRAPHS NO. 12

D I

17.0 17.5 I$.0 15.5 19.0 19.5 20.0 20.5 21.0

TARSAL LENGTH (ram)

FIGURE 31. Tarsal lengths of representative populations of Zosterops borbonica on Reunion Island. A. Localities II, 12, and 14 combined: 27 males and females; B. Localities 44 and 51 combined: 28 males and females; C. Locality 58: (I) 23 brown morphs, (2) 12 gray morphs; D. Localities 41 and 73 (26 gray morphs only). of white-eye. The distancesinvolved in clinal aspectsof the variation,as well as for the situation as a whole, are less than for any other example of geographicvariation in the family.

TAXONOMYOF Zosterops borbonica Variationin sizeand color of underpartsprovide no adequatebases for taxonomicdistinction of geographicallyseparated populations of Zosterops borbonicaon ReunionIsland. No usefulpurpose would be servedby formal recognitionof the endsof the ecogeographicdines in thesecharacters. This appliesin particularto Z. b. alopekionStorer and Gill, whichis simplythe darkbrown-breasted extreme of underpartcolor variation in somebrown morphs;it is the predominantform at only a few inlandlocalities. In addition, continuedrecognition of the gray morph,Z. b. edwardnewtoni,as restricted by Storerand Gill (1966) andStorer (1968), is clearlyinappropriate because it occurscommonly with the brown morph. Various argumentsexist for the continuedrecognition of a brown-headedbrown-backed form, Z. b. xerophila Storerand Gill (whichincludes Z. b. alopekionStorer and Gill), and a gray- headedbrown-backed form, Z. b. borbonica(Gmelin). These are geo- graphicallydistinct semispecies,but argumentsfor their recognitionare weakenedby the presenceof an inallocatablecolor phase (the gray morph) commonto both populations,and by the presenceof "brown-naped"individuals,which althoughthey form a homogeneouscoastal population, are also found as variants in the main populationsof both "borbonica" and "xerophila." In my opinion no combinationof existingor additionalsub- 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 49 specieswill adequatelyreflect thesecomplex patterns of geographicvariation and their probable evolution, and at the same time conform to current standardsof taxonomicvalidity. Consequently,the formal classificationof Moteau (1967) shouldbe followed (i.e., all Reunion Island forms shouldbe referredto as Z. b. borbonica)and a separateeffort be made to understand the variation in Zosterops borbonica on Reunion Island.

RELATION OF Zosterops borbonica mauritiana TO THE REUNION ISLAND FORMS

Zosteropsborbonica mauritiana of MauritiusIsland is a gray-backed,mono- morphicform of Zosteropsborbonica. There is no evidenceof geographic color variation in this form (Storer and Gill, 1966:7). In plumagecolor it is basicallya gray morph lacking phaeomelaninsexcept for a slight and variable brown wash on the flanks and occasional traces of brown on the back. However,the white underpartsof Z. b. mauritianadistinguish it from nearly all graymorph individuals on ReunionIsland; only a few extremepale-bellied gray birds from Reunion Island approachmauritiana in appearance.Z. b. mauritiana also differs in vocalizationsfrom Reunion Island borbonica (Gill, 1971).

EVOLUTIONARYHISTORY OF Zosterops borbonica Here I proposea probablesequence of historicalevents and evolutionary changesthat have led to the presentday pattern of variation and distribution of Zosteropsborbonica. It seemsclear to me (Gill, 1971 ) that the ancestors of borbonica colonized the Mascarene Islands after the ancestors of Z. olivacea;i.e., borbonicawas the secondsuccessful colonist in this particular case of double invasion of an oceanic island. Also Z. borbonica is almost certainlyderived from a generalizedwhite-eye tha,t favored edge, disturbed, or relativelyopen habitats; it has changedlittle from theseancestral habits. Before the arrival of man, habitatssuitable for Z. borbonicawere, in my opinion, rather limited on Reunion and existed only in relatively isolated pockets,such as the dry savannaof the westcoast and the highlandheath zone. Denseevergreen forest once extended to the seaon the wet eastcoast (Rivals, 1952:64) so that throughoutthis region Z. borbonicawas probably restrictedto the streamsidesand semi-openravines and perhapscoastal border of the forest,much as it is today. Small numbersof borbonicamay have pene- trated the island's interior along major streambeds,but the specieswas probablyabsent from the evergreenforest that coveredmost of the island. Erosion and landslidesmust have created some satisfactoryedge habitat in the cirques,and in both Cilaosand Marate this was probablycontinuous with the savannathat penetratedfrom the west coast via the ravine slopes 50 ORNITHOLOGICAL MONOGRAPHS NO. 12

(Rivals, 1952:50). Finally the forests below le Volcan were certainly damagedregularly and renderedsuitable by eruptionsand new lava flows. Thus severalpockets of habitatpartially isolated by stretchesof uninhabitable evergreenforest were originallyavailable to Zosteropsborbonica on Reunion Island. Zosteropsborbonica's ances•tor was very likely a typical yellow-green white-eyewith limited synthesisand depositionof phaeomelaninpigments in the flank feathers. Loss of carotenoidsin the plumage and lossof the eye- ring, which may have facilitatedcoexistence with the aggressiveZ. olivacea (Gill, 1971:54), may have resultedin a gray white-eyethat resembledthe presentday Z. b. mauritianain predominantcoloration including the brownish washon the flanks. That the original form of Zosteropsborbonica was more like a modern-daygray morph than a brown morph is suggestedby the rarity of extensivephaeomelanin pigmentation in Zosteropidaeand also by the presenceof a monomorphicgray form on Mauritius Island. Sometime after Z. borbonica lost its carotenoids and was established on both Mauritiusand Reunionislands, a phaeomelanicmorph probably evolved in the Reunion lowlandsand, through some selectiveadvantage at low altitudes, it gradually replaced gray morphs throughout the lowlands. Gray morphsthus becamerestricted to the disjuncthighland heath zone where they becameincreasingly adapted to the cold temperaturesand to utilizationof the scrub-like vegetation. I envisionthe subsequentdifferentiation within the brown morph as occurring in semi-isolatedpockets of suitable lowland habitat: the brown- headedform in the savannaof the island'swest side; the gray-headedform in the coastalvegetation and along stream beds on the east side; and the brown-napedform in the disturbedyoung coastalforests below Piton de la Fournaise,although the last may have arisen through hybridizationbetween the other two forms. It is impossibleto be certainof the extentand natureof the contactsbetween these populationsbefore the arrival of man on Reunion Island in 1663, althoughthe contactsmust have been limited in extent. There is little doubt that human activities,including lumberingand burning of the forests from the seventeenthto nineteenthcenturies, greatly changedthe aspect of the island'svegetation and openedmuch of the previouslyuninhabitable forest to Zosteropsborbonica. This wouldhave been true throughoutthe lowlandsand especiallyon the westernslopes of the island. In the latter regionmuch of the forest was quickly destroyed,the savanna and highland Acacia forests for their valuabletimber, and the evergreenforest in connectionwith the com- mercial cultivationof geranium. Damage was less extensivein the rugged, largely inaccessible,and very wet forestsof the easternslopes. As a result the gray-headedbrown morphson the east side of the island 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 51 were no longerrestricted to the rather limited edgehabitat alongstream beds and ravines; they increasedin numbers and expanded toward the savarma on the northeastwhere they cameinto contactwith the brown-headedbrown morphs. Eventually,gray-headed brown morphsadvanced westward as far as the Rivi•re des Galets where the combinationof aridity and the river bed provedto be an effectivebarrier betweenthe two forms. The "brown-naped" form expandedsimilarly along the southerncoast to the Rivi•re St. Etienne. The originalcontact of the "brown-naped"population and the populationof gray-headedbrown morphswas probably very similar to that found today, beingrestricted by new lava flows southof Bois Blanc. Brown morphsalso expandedtheir altitudinalrange and, particularly on the clearedwest slopes,met the gray morphswhich were beginningto occupy lower elevationsthan was previouslypossible. Inasmuch as gray morphswere adaptedto high altitudesand brown morphslow .altitudes,a cline in the proportions of the two was established.The expansionof brown morphs into thehighlands also led to the.evolution of a seriesof populationsadapted in size and pigmentationto local conditions. The rather rapid evolution of these differenceswithin 250 years is in keepingwith recent evidencefor rates of racial differentiationin the House Sparrow (Passer domesticus) (Johnston and Selander, 1964; Packard, 1967).

CHARACTER VARIATION AND ADAPTATION The existenceof infraspecificgradients of pigmentationand size that are correlated with climate has been recognizedsince the earliest studies of geographicvariation, when it led to the formulationof "ecogeographicrules." More recent studies (James, 1970; Power, 1969; Johnston and Selander, 1964, 1971) havedemonstrated the precisionand rapidityof suchevolutionary adaptation. The selectiveadvantage of variation in size with relation to temperature differencesis presumedto be physiologicaland related to thermoregulation. This hasbeen disputed (Scholander, 1955), but recentevidence (Brown and Lee, 1969) supportsMayr's intuitive arguments(1956) for the existenceof adaptivedifferences in bodysize. The relationof relativehumidity to thermo- regulationand consequentlyits effectson bodysize have recently been recognized (James, 1.970). The correlation of size in some Zosterops borbonica with altitude,as well as in other bird species(Rand, 1936), presumablyalso reflectsthermoregulatory adaptation to temperatureand humidity. The mean- ing of inversecorrelations of size with altitudeas found in the high altitude gray morph is not known. Finally in additionto being correlatedwith body size,wing length is subjectto a varietyof otherselective forces including the effectsof altitudinalchanges in air pressureon flight capability (Hamilton, 1961:185; Moreau, 1957:331). 52 ORNITHOLOGICAL MONOGRAPHS NO. 12

The evolutionof colordifferences as protectionfrom visualpredators is well known. However, differential loss of the color forms of Z. borbonica to visual predatorsdoes not seemto be significanton ReunionIsland, if it occursat all. Most of the predatorson the island,e.g., rats, lizards, snakes, cats, and humans, are restricted in their distribution and are recent arrivals on the island. The one widespreadpredator againstwhich some protectivecoloration would seemto be advantageousis the harrier, Circusaeruginosus, but this hawk does not posemuch of a threatto passerineslike Zosteropsborbonica (see p. 12). Recent evidence (Lustick, 1969, 1971; Hamilton and Heppner, 1967; Heppner,1970) indicatesthat the absorptionof solarenergy by plumagepig- mentsis importantin avianthermoregulation, particularly because of its effects on the insulationvalue of plumage. Thus increased(melanin) pigmentation and thereby heat absorptionmay be advantageousin cooler climates. The altitudinalgradients in underpartcoloration of Zosteropsborbonica, which often face the sun on exposedperches early in the morningat high altitudes, might be accountedfor in this way. Increasedmelanin depositioncould also help maintainthe insulationvalue of the plumageby reducingfeather wear, as well as by assistingin the drying of feathersthrough the absorptionof radiant energy. Although speculativeonly, this latter hypothesiscould help explain the fact that many birds have increasedmelanin depositionin cold humid climates,i.e., "Gloger'sRule." At presentthere are no satisfactoryexplanations to accountfor the differences (i.e., brown vs gray) in back colorationor head coloration of Z. borbonica. Physiologicalattributes other than plumage color per se seem importantin determiningthe geographicaldistributions of theseforms.

CHARACTER DIVERGENCE IN NATURAL POPULATIONS

The principal biological interest in the variation in Reunion Island forms of Zosteropsborbonica is that sucha smallgeographic area is involved. Such marked microgeographicvariation in birds is not only unknown on other oceanicislands, but is alsoexceptionally rare on larger land masses. There is no reason to assume that Z. borbonica is an unusual or remarkable bird species.Behaviorally it is a typical,social white-eye with generalized feedinghabits, and it doesnot appearto be a uniquelysedentary bird species. Locally it is highlymobile, habitually flying considerabledistances over the forests,and it is no way restrictedto the groundor denseunderstory. I have observedbanded individuals up to 500 metersfrom the point at whichthey were initially capturedand banded,and I suspectthat they wander freely within severalsquare kilometers of suitablevegetation, perhaps much more. Generallyit seemsas if thehigh dispersal tendencies of birdsand the gradual nature of environmentalgradients preclude highly localized adaptation. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 53

Furthermore,the importanceof geographicisolation, or disruptionof the continuityof populations,seems undeniable as a factorpromoting evolutionary divergence(Mayr, 1970:296-330). However, the probability and rate of geneticdivergence depends also on the magnitudeof the selectioncoefficients in the different environments.The relationshipbetween these two factors must be examined to ascertain the basis of differences or the lack thereof betweenpopulations. Direct measuresof selectioncoefficients in natural populations,based on temporal fluctuationsin the proportionsof genotypes(Clarke and Murray, 1962; Gershenson,1945; Merrell and Rodell, 1968), relativeviability (Ford, 1964:50, 62; Sheppardand Cook, 1962; Fisher, 1939), or differentialcapture by predators(Kettlewell, 1961a, 1961b), are availablefor only a few well studiedpopulations of moths, butterflies,snails, and small mammals. The selectioncoefficients indicated by thesestudies range from 5 to as high a.s90 per centrelative advantage of one genotypeover another. In two studiesthe selectioncoefficients required to offsetthe effectsof dis- persalin natural casesof clinal variation have been computed;the computed selectioncoefficients were very low -- 0.1 per cent in PeromyscuspMionotus (Haldane, 1948:284) and .004 per cent in Amathesglareoxa (Kettlewell and Berry, 1961) -- and werewell belowthe actualintensities of selectionthought to be operating. The mathematicalrelationships between natural selection and dispersal (= gene flow) and the distancesover which these two forces interact to produceclinal variation have been treated by Haldane (1948) and Fisher (1950) for simplegenetic systems. Haldane's method assumesthe following: 1) the specieslives in an area which is plane and infinite, i.e., it must be large enoughthat regionsexist where there is no polymorphism;2) the density is equal throughoutthe area; 3) the animalsdisperse at random; 4) they have one annual generation; 5) mating is at random betweenthe different types; 6) the populationis in equilibrium; 7) an autosomaldominant and its allelomorphunderlie the polymorphism; and 8) the plane is sharply dividedby a straightboundary into two halves. Assumptions1 thru 6 are acceptablewithout major violationfor Zosterops borbonica. Assumption5 may not be completelyvalid, however. The presenceof helpersand communal feeding of youngin the nestmight be taken as indirect evidence of inbreeding, as such behavior would be selectively advantageousto the participatingindividuals if the parentsand helperswere relatedgenetically (Lack, 1968:80). Allochronicdifferences in the onsetof breedingmay reduce altitudinalgene flow even further and reinforcethe geographicisolation between lowland populations. Finally, any tendency toward differential habitat selection, for which there is some evidence (see p. 17), will alsoreduce gene flow. 54 ORNITHOLOGICAL MONOGRAPHS NO. 12

The underlyinggenetic bases of the polymorphismin this white-eye (As- sumption7) are not clear. However,other similar cases of colorpolymorphism in birds (Cooke and Cooch, 1968:299; Munro et al., 1968; Hrubant, 1955: 229) appearto be controlledby two alleleswith one being dominantand with modifiersbeing responsible for the intermediateplumage types. The re- quirementof a sharpboundary (Assumption 8) is violatedby our situation. However,Haldane (1948:284) recognizesthat in many casesenvironmental conditionsdo changegradually and he thinksthat in thesecases the computed selectioncoefficient should at leastindicate the order of magnitudeof selection neededto maintain a cline, i.e., whether selectiveintensities of the order of 10 or .01 per cent are involved. The measurementsinvolved in Haldane'sformula (1948:283 ) are the inter- quartiledistance (d), which is the linear distancebetween the two.localities at whichthe frequenciesof one of the phenotypesare 25 and 75 per cent,and the averagedispersal distance (m) from hatchingsite to first breedingsite. The intensityof selectionin the interquartileregion is then consideredto be m2/2d 2. There are no directmeasures of dispersaldistances for Zosteropsborbonica. Mean distancesof dispersalhave, however, been measuredfor .severalocher bird species.The Song Sparrow (Melospiza melodia) dispersesabout 400 meters (Johnston, 1961). The Wrentit (Chamaea fasciata), an inhabitant of densebrush and chaparral,might disperseabout 600 to 700 meters (Miller, 1947). The mean dispersaldistance of the Great Tit (Parus major) is also about 600 meters (Kluijver, 1951: 14). I would estimate,therefore, that the meandispersal distance in Zosteropsborbonica is about500 meters. However, the highly social nature of borbonica'syear-round activities,including the apparentlack of territoriality,must promotethe formationof local and per- hapsinbred populations. In order to compensatefor this, as well as for the earlier statedpossible violations of Assumption5, I have consideredthat the effectivedispersal distance might be as little as 250 meters. This seemslike an absoluteand probably exaggerated minimum possibility for a highlymobile bird. At the other extreme, I doubt that Zosterops borbonicais much more mobile than someof the other speciesmentioned above, and would set 1,000 metersas the upperlimit to its mean dispersaldistance. UsingHaldane's formula, then, I have estimatedthe interquartileselection coefficientsfor four of the five clinesin the ratio of brown and gray morphs whichI sampledon ReunionIsland. Separatevalues were computedassuming mean dispersaldistances of 250, 500, and 1,000 meters (Table 12). The resultingselection coefficients range from a low of 0.06 per centto a maximum oœ5.5 per cent. These calculated selection coefficients are not intended to be actual measures of selectionon ReunionIsland, but they do provideestimates of the approxi- 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 55

TABLE 12 INTENSITY OF SELECTION REQUIRED FOR MAINTENANCE OF CLINES IN THE RATIO OF GR•Y AND BROWN MORPHS 1N Zosterops borbonica

Calculation Transect •

A B C D

Locality number with 25 per cent gray morphs 33 6 74 62 Locality number with 75 per cent gray morphs 41 5 71-73 64 Distance (d) between the two localities (in km) 7 4 3 4 Selection coefficient 2 With m---- 250 meters .06 .19 .35 .19 With m---- 500 meters .25 .78 1.40 .78 With m = 1,000 meters 1.02 3.10 5.50 3.10

• A, St. Benoit to Plaine des Carres; B, St. Denis to Plaine des Chicots (locality 6 is adjacent to but not actually on this transect); C, Petite France to Piton des Epinards; D, gtang Sa16 les Bains to Piton des Epinards (see Figure 11). •The selection coefficient (K) =m•/2d •, where m is the assumed mean dispersal distance and where d is the distance as defined above (after Haldane, 1948). Values indicated are per cent relative advantage of a morph on that part of the cline where it comprises more than half of the population. mate minimummagnitude of selectionthat seemsto be operating. From them it is apparent.that a small selectiveadvantage is sufficientto maintain steepclinal variation in simplemorph or allelic frequenciesin a mobile small bird. Studiesof natural populationsincreasingly are revealingdemal structures and selectioncoefficients capable of evolvingmarked local populationdiffer- encesdespite high dispersalrates (Ehrlich and Raven, 1969). There are, of course,lower limits to the size of a geographicalarea within which a given speciescan differentiate,but theseare well below the actualranges for most species.Also, geneticmarkers reveal differences that are muchmore localized than are the differencesin external morphologywhich classicallyhave been usedto describegeographic variation. It is importantto recognize,therefore, that geneflow per se (i.e., the resultingtendency towards genetic similarity) has probablybeen overemphasizedas a factor preventingthe divergenceof conspecificpopulations. Rather, character uniformity over large areas probably reflects a generalizedphenotype that adaptively compromisesa varietyof opposingselective forces. Characteruniformity may alsoreflect a species'inability to evolve locally adapted populationsbecause of genetic homeostasis(see Mayr, 1963:289). In strong contrast to the clinal situationsfound in the highland areas on Reunion Island, in the lowlandscolor forms of Zosteropsborbonica change abruptly acrossobvious geographical barriers (river beds and a lava flow). 56 ORNITHOLOGICAL MONOGRAPHS NO. 12

For example, the coas{alranges of gray-headedand brown-headedbrown morphsare separatedby the gravelbeds of the Rivibre des Galets and the Rivibre St. Etienne, but the transition between these two forms above 1,400 meterson the eastside of ReunionIsland is steeplyclinal without any apparent barrier. In fact, all altitudinal transitions between color forms are clinal, whereas all coastal contacts involve a barrier. Furthermore, the coastal transitionstake place in the absenceof conspicuousenvironmental change. One might conclude,therefore, that the selectioncoefficients affecting the altitudinaldistribution of genotypesare large enoughto maintaingeographic differenceswithout the restrictionof gene flow, whereasthe selectionco- efficientsin the lowlandsare lower. Geographictransitions between genotypes in the lowlands,therefore, require some restrictionof gene flow. A delicatebalance between gene flow and opposingselection forces must be involved. Finally, in analyzingthe forces affectingthese characterdivergences, the nature of Reunion Island itself must be considered. Few small islands are as high as ReunionIsland or containcomparable ranges of environmentalex- trem.eswithin their confines. Furthermore, a tropical or subtropicallocation maximizesclimatic diversitybecause a very high island can include both tropical lowlandsand alpine mountain{ops. Few small tropical islands,but several subtropicalislands, including Reunion, Hawaii, Maul, Bali, and Lombok, are that high. Characterdivergence in birds on theseother islands shouldbe lookedfor. The essentialabsence of visualpredators from Reunion Island may alsohave been an importantfactor in the evolutionof Zosterops borbonica,since release from a need for protectivecoloration may have in- creasedthe evolutionaryflexibility of borbonica'splumage coloration. Taken together these considerationssuggest that the marked differentiation in Zosteropsborbonica on ReunionIsland resulted from and was made possible by severalfactors which in themselvesare not unusualbut which are rarely found in combination in such a small area.

SUMMARY Zosteropsborbonica, a white-eyeendemic to the MascareneIslands in the westernIndian Ocean,is characterizedby complexpatterns of plumagecolor and size variation within the confinesof Reunion Island, a remote volcanic islandabout 1,000 squaremiles in area and 3,000 metershigh. Z. borbonica appearsto be the secondof the two indigenouswhite-eyes to colonizethis island and has lost the carotenoidpigmentation and white eye-ring characteristicof most of its relatives. However, it is still a typicallysocial white- eyethat has a generalizeddiet and preferencefor edgeand disturbedhabitats. One aspectof the colorvariation in Z. borbonicainvolves the presenceo.r absenceof phaeomelaninsin the feather barbs of the back plumage,which 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 57 producesbrown and gray colormorphs, respectively. The offspringof mixed matingsbetween morphs were observed; the geneticbasis of backcolor seems to involvea simplepair of alleleswith somemodifiers. Thesetwo color forms are extensivelysympatric, but the ratio of the two varies clinally with altitude from 100 per cent brownmorph populations occurring in the lowlandsto predominatelygray morph populationsoccurring in the highlands.Linear distancesbetween localities at whichgray morphs composed 25 per centand 75 per centof thepopulation varied from threeto sevenkilometers. The minimum intensityof selectionrequired to maintainsuch steep altitudinal clines in the proportionsof the two morphsis estimatedto be on .theorder of 1 per centrelative advantage of onemorph over the other. A secondaspect of the color variation involvesdifferences in the head color of brownmorphs only. Threecategories of brownmorph head color variants can be distinguished:an all gray-headedform whichis characteristicof the wet northernand easternlowlands below 1,380 meterselevation; an all brown- headedform whichis foundon the dry westernslopes and in the highlands above1,400 meterselevation; and a brown-napedintermediate which is the form characteristic of the southern coast. Lowland contacts between these differentbrown morphs are abrupt,coinciding with the Rivigredes Galets (gray headsvs brownheads), the RivigreSt. Etienne (brown headsvs brown napes),and recent lava flowsnear Bois Blanc (brown napes vs grayheads). In contrastthe altitudinaltransition between the gray-headedand brown- headedforms on the easternslopes of ReunionIsland is continuousand steeplyclinal over a distanceof aboutfive km; samplesfrom this contactzone includea largenumber of purebrown- or gray-headedforms as well as inter- mediatephenotypes. The intensity of eumelanindeposition in the feathersof the underparts variesin relationto altitudein all colorforms, the underpartschanging from whitishin lowlandsto dark gray in the highlands.Extremes of this color variationare more strikinglydifferent than is the casein mostmainland sub- speciesof Zosteropswhich exhibit this variation. The highlysignificant correlationsof thisvariation with altitudeand the absenceof significantvisual preda- ters on ReunionIsland suggestthat the differenceshave physiologicalvalue in termsof absorptionof radiantenergy. In brown-headedbrown morphs the intensityand extent of phaeomelanindeposition on the underpartsalso increasesin relationto increasingaltitude and rainfall. Size characteristicsvary betweencolor forms and in relationto altitude. Wing length of brown morphs increasesabout one millimeter per 1,000 meters of elevation. Bill and tarsal lengths also increasewith altitude in brown-headedbrown morphs. Bill length decreaseswith altitude in gray morphs. Gray-headedbrown morphshave longer bills than lowland brown- headedbrown morphs, but aboutthe samelength bills as large highland brown 58 ORNITHOLOGICAL MONOGRAPHS NO. 12 morphs.Lowland gray-headed brown morphs and highlandgray morphs show sexualdimorphism in bill length. The following sequenceof possibleevolutionary events is envisionedas beingresponsible for the presentsituation. The brown morphfirst appeared as a mutant form in the originalgray morph populationsof Reunion Island. They graduallyreplaced gray morphsin the lowlandsbut not the highland heath zone and evolved head color and associated differences while isolated in disjunctpockets of suitable"edge" habitat, e.g., brown-headedbrown morphs in the savannaof the arid westcoast, gray-headed brown morphs along rivers and streambeds in the east,and brown-napedbrown morphs in youngforests at the baseof the activevolcano in the south. It is possiblethat the brown- napedbrown morph arose through hybridization between the othertwo forms. The presentday pattern of geographicdistribution of these color forms reflects range expansionsand secondarycontacts that resultedfrom human destructionof indigenousforests. Associatedaltitudinal range expansions havebrought brown and graymorphs together and clineshave developedre- fleetingthe relativeadaptation of the morphsto the lowlandsand highlands. The expansionsalso have led to the developmentof steepprimary clinesin the intensityof melanindeposition and sizewithin the past250 years. The highly localizeddifferentiation of populationsof Z. borbonicaappears to havebeen the inevitableresult of severalfactors, e.g., its socialand sedentary habits,which, however,are not exceptionalor even extreme,its ties to dis- turbedand edgehabitats, the climaticdiversity of ReunionIsland, and perhaps increasedadaptive flexibility resultingfrom releasefrom someselective agents usuallypresent on continents,such as visualpredators. In themselveseach of thesefactors is not very unusual,but they are rarely found in combination within the confines of a small oceanic island. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 59

LITERATURE CITED

BAILEY, N. T.J. 1959. Statistical methods in biology. London, The English Uni- versities Press Ltd. BArl)W•N,S. P., H. C. OBEVa-IorSER,ANl) L. G. WORrE¾. 1931. Measurementsof birds. Sci. Publs. Cleveland Mus. Nat. Hist., Vol. 2. BERrIOZ,J. 1946. Faune de l'empire franqais. IV. Oiseaux de la R6union. Paris, Librairie LaRose. BOWERS,D. E. 1956. A study of methods of color determination. Syst. Zool., 5: 147-160. BOWERS, D. E. 1960. Correlation of variation in the Wrentit with environmental gradients. Condor, 62: 91-120. BROWN,J. H., AND A. K. LEE. 1969. Bergmann'sRule and climatic adaptation in woodrats (Neotoma). Evolution, 23: 329-338. CHAMArAUN,F. H., Am) I. McDouanrr. 1966. Dating geomagnetic polarity epochs in R6union. Nature, 210: 1212-1214. CLAmrE,B., Am) J. MURRAY. 1962. Changes in gene-frequencyin Cepaea nemoralis (L.); The estimationof selectivevalues. Heredity, 17: 467-476. CooIcE, F., nNl) F. G. Coocu. 1968. The genetics of polymorphism in the goose Anser caerulescens. Evolution, 22: 289-300. EDDINGER,C. R. 1967. Feeding helpers among immature white-eyes. Condor, 69: 530-531. EHRLICH, P. R., AND P. H. RAVEN. 1969. Differentiation of populations. Science, 165: 1228-1232. FISHER,R.A. 1939. Selectiveforces in wild populationsof Paratettix texanus. Ann. Eugenics, 9: 109-122. FISHER,R.A. 1950. Gene frequenciesin a cline determinedby selectionand diffusion. Biometrics, 6: 353-361. Fore), E.B. 1964. Ecologicalgenetics. London, Methuen and Co., Ltd. FV,nNK, F. 1938. PigmentanalytischeUntersuchungen am Rassenkreis,Parus atricapil- lus L. Proc. 9th Intern. Ornith. Congress (Rouen): 161-175. GAl)OW,H. 1884. Catalogue of the birds in the British Museum, Vol. 9. London, Taylor and Francis. GERSHENSON,S. 1945. Evolutionary studies on the distribution and dynamics of melanism in the hamster (Cricetus cricetus L.). II. Seasonal and annual changes in the frequencyof black hamsters. Genetics,30: 233-251. GILL, F. B. 1971. Ecology and evolution of the sympatric Mascarene white-eyes, Zosteropsborbonica and Z. olivacea. Auk, 88: 35-60. HArl)ANE,J. B.S. 1948. The theory of a cline. J. Genetics, 48: 277-284. HAMILTON,T.H. 1961. The adaptive significancesof intraspecifictrends of variation in wing length and body size among bird species.Evolution, 15: 180-195. HAMILTON, W. J., III, AND F. H. HEPPNER. 1967. Black pigmentation: adaptation for concealment or heat conservation. Science, 158: 1341. HAm•IsoN, C. J.O. 1965. Allopreening as agonistic behaviour. Behaviour, 24: 161- 204. HARTLAUB,G. 1877. Die Voegel Madagascars und der benachbarten Inselgruppen. Halle, Drnck and Verlag yon H. W. Schmidt. HEPPNER,F. 1970. Metabolic significanceof differential absorptionof radiant energy by black and white birds. Condor, 72: 50-59. HRUBANT,H.E. 1955. An analysisof the color phasesof the eastern ScreechOwl, Otus asio, by the gene frequency method. Amer. Nat., 89: 223-230. 60 ORNITHOLOGICAL MONOGRAPHS NO. 12

JAMES,F.C. 1970. Geographicsize variation in birds and its relationshipto climate. Ecology, 51: 365-390. JOHNSON,N.K. 1966. Morphologic stability vs adaptive variation in the Hammond's Flycatcher. Auk, 83: 179-200. JOHNSTON,R.F. 1961. Population movements of birds. Condor, 63: 386-389. JOHNSTON,R. F., AND R. K. SELANDER.1964. House Sparrows: rapid evolution of races in North America. Science, 144: 548-550. JOHNSTON,R. F., AND R. K. SELANDER. 1971. Evolution in the House Sparrow. II. Adaptive differentiation in North American populations. Evolution, 25: 1-28. KETTLEWELL, H. B. D. 1961a. Selection experiments on melanism in Amathes glareosa Esp. (Lepidoptera). Heredity, 16: 415-434. KETTLEWELL,H. B. D. 1961b. The phenomenon of industrial melanism in Lepi- doptera. Ann. Rev. Entomol., 6: 245-262. KETTLEWELL,H. B. D., aND R. J. BERRY. 1961. The study of a cline. Heredity, 16: 403-414. KLUIJVER,H.N. 1951. The population ecologyof the Great Tit, Parus rn. major L. Ardea, 39: 1-135. LACK,D. 1968. Ecological adaptations for breeding in birds. London, Methuen and Co., Ltd. LUSTICK,S. 1969. Bird energetics: effects of artificial radiation. Science, 163: 387- 390. LUSTInK, S. 1971. Plumage color and energetics. Condor, 73: 121-122. MAY•, E. 1956. Geographicalcharacter gradients and climatic adaptation. Evolution, 10: 105-108. MAY•, E. 1963. Animal species and evolution. Cambridge, Belknap Press. MAY•, E. 1970. Population, species,and evolution. Cambridge, Belknap Press. MERRELL, D. J., AND C. F. RODELL. 1968. Seasonal selection in the leopard frog, Rana pipiens. Evolution, 22: 284-288. MILLER, A. H. 1941. Speciationin the avian genus Junco. Univ. California Publs. Zool., 44: 173-434. MILLER,A.H. 1947. Panmixia and populationsize with referenceto birds. Evolution, 1: 186-190. MILON, P. 1951. Notes sur l'avifaune actuelle de File de la R6union. Terre et la Vie, 98: 129-177. MOREAU,R. E. 1957. Variation in the western Zosteropidae (Aves). Bull. British Mus. (Nat. Hist.), Zoology, 4: 312-433. MoaEAu, R. E. 1967. [Family Zosteropidae, African and Indian Ocean taxa.] Pp. 326-337 in Check-list of birds of the world, vol. 12 (R. A. Paynter, Jr., Ed.). Cambridge, Massachusetts,Mus. Comp. Zool. MUNRO, R. E., L. T. SMITH, AND J. J. KUPA. 1968. The genetic basis of color dif- ferencesobserved in the Mute Swan (Cygnus olor). Auk, 85: 504-505. NEWTON,A., ANDE. NEWTON. 1888. Notes on some speciesof Zosterops. Ibis (Fifth Series) 6: 474-476. PAci•nm), G. C. 1967. House Sparrows: evolution of populations from the Great Plains and Colorado Rockies. Syst. Zool., 16: 73-89. POLLEN, F. P. L., AND D.C. VAN DAM. 1868. Recherches sur la Faune de Mada- gascar et de ses d6pendences,d'apres les decouvertesde F. P. L. Pollen and D.C. van Dam. Pt. 2. Mammiferes et Oiseaux. Leiden. POWER,D.M. 1969. Evolutionary implicationsof wing and size variation in the Red- winged Blackbird in relation to geographicand climatic factors: a multiple re- gressionanalysis. Syst. Zool., 18: 363-373. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 61

RAND, A.L. 1936. Results of the Archbold Expedition No. 12. Altitudinal variation in New Guinea birds. American Mus. Novit., no. 890. RIVALS,P. 1952. Etudes sur la vegetation naturelie de l'ile de la R6union. Travaux Laboratoire Forestier. Toulouse. Tome V, Geographie forestiere du monde, Vol. I, Art. II. SCHOLA•qDER,P.F. 1955. Evolution of climatic adaptation in homeotherms. Evolu- tion, 9: 15-26. SHEPPARD,P.M., AND L. M. COOK. 1962. The manifold effects of the medionigra gene on the moth Panaxia dominula and the maintenanceof a polymorphism. Heredity, 17: 415-426. Sm6EL, S. 1956. Nonparametric statisticsfor the behavioral sciences.New York, McGraw-Hill Book Co. STORER,R.W. 1968. The cotypesof ZosteropsE. Newtoni Hartlaub. Auk, 85: 127. STOR•R, R. W., AND F. B. GILL. 1966. A revision of the Mascarene white-eye Zosterops borbonica (Aves). Occ. Papers Mus. Zool. Univ. Michigan No. 648. UPTON, B. G. J., ANDW. J. WADSWORTH. 1966. The basalts of Reunion Island, Indian Ocean. Bulletin Volcanologique, 29: 7-24. WATSON,G. E., R. L. Zus•, ANt) R. W. STOR•R. 1963. Preliminary field guide to the birds of the Indian Ocean. Washington, Smithsonian Inst. 62 ORNITHOLOGICAL MONOGRAPHS NO. 12

APPENDIX I SUMMARY OF REUNION ISLAND LOCALITIES 1

Locality Altitude Location Name Number of number (meters) specimens

1 1,000 2 km SSE le Brill6 Val Fleuri 14 2 1,100 3 km SSE le Brill6 Mamode Camp 9 3 1,580 5.7 km SSE le Brill6 14 4 1,780 7 km S le Brill6 Plaine des Chicots (gite) 3 5 1,910 7.7 km S le Brill6 Plaine des Chicots 6 6 1,650 5.2 km E Dos d'Ane Plaine d'Affouche 9 7 1,460 3.3 km E Dos d'Ane Plaine d'Affouche 0 8 1,700 20 km NNE St. Joseph Rav. Rivi•re 8 des Remparts 9 1,000 8 km SSW St. Denis Chemin Arnoux 11 10 820 5.3 km ESE la Possession Piton d'Orange 13 11 90 2 km S St. Denis Ravine du Butor 10 12 220 3.5 km S Ste. Marie Beaufonds 17 13 10 2.7 km E Ste. Marie Ravine Chevres 0 14 200 5.2 km NE Salazie Bras de Caveme 9 15 560 1.5 km SSE Salazie Salazie 0 16 680 2 km ESE Gd. Islet Rivi•re Fleurs Jaunes 13 17 1,100 1 km NE Gd. Islet Grand Islet 4 18 1,530 3 km SSW Gd. Islet Col Bemale 11 19 1,940 7 km W Hellbourg Col de Fourche 11 20 1,160 4 km W Hellbourg Ravine Farla 10 21 920 2 km W Hellbourg Rivibre du Mat 3 22 1,350 1 km S Hellbourg Terre Plate 6 23 1,570 2.2 km SW Hellbourg Plateau de Belouve 10 24 1,400 3.8 km NE Hellbourg Trou de Fer 10 25 640 9 km SW St. Benoit Ravine Mathurin 11 26 1,010 12 km SW St. Benoit Takamaka 11 27 1,180 5 km WNW P1. des Plateau Duverney 10 Palmistes (1 er) 28 1,360 7.5 km W P1. des Forgt du B6bour 0 Palmistes (1 er) 29 1,350 6.5 km WNW P1. des Forgt du B6bour 15 Palmistes (1 er) 30 1,720 9.2 km W P1. des Plaine des Salazes 10 Palmistes ( 1er ) 31 720 4 km NE P1. des Bras Grand Piton 11 Palmistes (1 er) 32 860 2.7 km NE P1. des Piton Camp de Tgte 1 Palmistes (1 •r) 33 1,060 1.5 km SSW P1. des P1. des Palmistes (2 eme) 6 Palmistes (1 or) 34 170 3.5 km NNE St. Joseph Rivi•re des Remparts 10

Palmistes 4 36 1,250 4.7 km SW P1. des 351,250 4.2PalmisteskmSW P1. des 37 1,620 6 km SE P1. des Rampe de la 0 Palmistes Grande Mont6e 1973 GILL: VARIATION IN ZOSTEROPS BORBONICA 63

APPENDIX I (continued)

Locality Altitude Location Name Number of number (meters) specimens 38 1,600 5.5 km le Vingt- Plaine des Carres 2 Septi•me 39 2,300 1 km ENE Pas de Pas de Bellecombe 8 Bellecombe 40 1,400 6 km WNW P1. des Piton de la Plaine 7 Palmistes des Cafres 41 2,060 4.5 km E le Vingt- Nez de Boeuf 16 Septi•me 42 2,300 7.5 km ENE le Vingt- Piton de Sable 0 Septi•me 43 2,300 8 km SSE le Vingt- Plaine des Remparts 0 Septi•me 44 0 2.5 km WSW St. Louis lgtang du Gol 13 45 520 5 km WSW Ste. Rose ForSt Mourouvin 12 46 500 4.2 km S Ste. Rose Forgt Mourouvin 11 47 60 3 km SSW Bois Blanc Grand Brill6 12 48 100 10.2 km SSW Bols Blanc Grand Brill6 11 49 1,060 17 km N St. Joseph Rivi•re des Remparts 12 50 20 3 km ESE St. Pierre Ravine des Cafres 0 51 40 1 km NE l'Etang Forgt Dominale 11 Sa16 les Bains 52 1,380 3.5 km NNE Plaine Bols Court 0 des Cafres 53 270 8.7 km NNE St. Louis Petite Serr6 7 54 340 5 km N St. Louis Bras de Patate 4 55 300 6.5 km SSE Cilaos Le Pavillon 11 56 1,000 2 km SSE Cilaos Bras de Benjoin 9 57 1,380 4.5 km S le Brill6 10 58 1,400 1 km N Cilaos ForSt du Gd. Matarum 37 59 2,000 3 km NNE Cilaos Petit Matarum 0 60 2,460 4 km NE Cilaos Cavern Dufour 10 61 1,080 1 km N T6velave For•t Soumise des 10 Bernards (PK 2) 62 1,440 2.5 km NNE T•velave ForSt Soumise des 21 Bernards (PK 7) 63 1,700 4 km NNE T•velave For•t Soumise des 29 Bernards (PK 10+) 64 1,980 5.7 km NNE T•velave Ravine des Avirons 8 65 180 3 km WSW Trois Bassins Ravine Chauve Souris 3 66 10 2.5 km S St. Gilles 0 67 0 2.5 km NNE St. Paul l•tang St. Paul 0 68 120 5.5 km SE le Port Rivi•re des Galets 10 (south side) 69 160 7 km SE le Port Rivigre des Galets 3 (north side) 70 1,660 11 km ESE St. Paul Ligne Dominiale 7 71 1,800 11.7 km ESE St. Paul Camp Dennemont 5 72 1,400 9.5 km ESE St. Paul (PK 5) 10 73 2,200 14.2 km ESE St. Paul Piton Maido 14 74 1,550 10.5 km ESE St. Paul (PK 7) 13 75 20 7 km WSW St. Paul Ravine Jacques 0 64 ORNITHOLOGICAL MONOGRAPHS NO. 12

APPENDIX I (continued)

Locality Altitude Location Name Number of number (meters) specimens

76 20 8 km WSW St. Denis Rav. de Grande 0 Chaloupe 77 20 !km NE la Possession Camp Magloire 0 78 40 I km SSE la Possession Grande Ravine 0 79 900 1.7 km W Cilaos Cilaos (Bras Rouge) 0 80 500 9.5 km SW St. Benoit Grand Etang 0 81 30 3 km SSE St. Benoit Ravine S&he 11 82 10 .5 km SSE Ste. Anne Ste. Anne 0 83 0 Ste. Rose Monument 0 84 20 5.2 km SE Ste. Rose Piton $te. Rose 0 85 20 2 km NNE Bois Blanc Pointe Cascades 10 86 500 6.5 km W St. Philippe Brfil• du Baril 6 87 70 4 km SE le Port Jardin Boyer 16 88 I0 2.5 km$E St. Philippe Mare Longue 0 89 1,360 4.6 km NE Dos d'Ane Plaine d'Affouche 0 90 680 4.7 km NE P1. des Morne de l'Etang 0 Palmistes (! •') 91 10 1.7 km W St. Pierre St. Pierre (Oratorio) 0 92 20 4.5 km WNW St. Pierre 0 93 90 4.2 km E St. Louis Riv. St. Etienne 11 (south side) 94 300 1.5 km NW les Avirons Ravine des Avirons 1! 95 640 1.7 km W T•velave Ravine des Avirons I0 96 740 I km SE le Brfil• Ravine Butor 10 (Le Brill6) 97 1,080 7.6 km ESE St. Paul La Petite France 10 98 1,330 1.8 km NNE T•velave For•t Souraise des I0 Bernards (PK 5.5) 99 380 6.6 km SW St. Benoit Chemin de la I1 Grand Fond I00 1,740 5 km NNW le Vingt- Plaine des Carres 6 Sepfi•me I01 1,940 7.7 km NNE le Vingt- Coteau Maigre 10 Septibme 102 280 5 km E Ste. Rose Riv. de l'Est !

All specimens were taken in 1967, except those for localities I00 and I01, which were taken in 1964. 1973 GILL: VARIATION IN ZOSTEROPS BORBON1CA 65

APPENDIX II UNDERPARTGRAY COLOR VARIATION IN Zosterops borbonica ON REUNION ISLAND

Locality ARitude Sample size Underpart gray number (meters) Mean index Standard deviation

1 1,000 14 4.6 1.50 2 1,100 6 4.8 1.83 3 1,580 11 6.5 1.13 6 1,650 9 5.8 0.45 8 1,700 8 6.2 1.75 9 1,000 11 3.1 1.37 10 820 11 3.3 1.19 11 90 10 2.7 1.25 12 220 13 2.5 0.87 14 200 9 4.0 1.22 16 680 13 3.9 1.44 18 1,530 11 5.5 1.44 19 1,940 10 6.0 1.05 20 1,160 10 6.2 1.31 23 1,570 10 5.3 1.42 24 1,400 10 5.9 1.41 25 640 8 4.1 1.25 26 1,010 11 5.1 1.30 27 1,180 10 5.3 0.67 29 1,350 10 5.5 1.35 30 1,720 9 6.5 0.84 31 720 11 5.0 1.79 33 1,030 6 5.5 1.76 34 170 10 1.1 0.73 39 2,300 6 6.3 1.30 40 1,400 7 6.4 0.78 41 2,060 24 5.6 1.10 44 0 13 1.7 0.87 45 520 12 3.3 1.37 46 500 11 4.4 1.21 47 80 12 2.7 1.37 48 100 11 2.4 1.86 49 1,060 12 4.8 1.75 51 40 17 2.4 0.87 53 270 7 2.1 0.69 55 300 7* 2.3 0.72 56 1,900 7 4.6 0.53 57 1,380 8 5.4 1.06 58 1,400 38 6.0 1.37 60 2,460 10 6.5 1.08 61 1,080 8 2.6 1.30 62 1,440 15'* 3.5 1.12 63 1,700 25 5.4 1.34 64 1,980 8 5.7 1.83 68 120 10 1.4 0.71 70 1,660 7 4.4 1.94 72 1,400 9 4.5 1.34 73 2,200 14 6.3 1.24 66 ORNITHOLOGICAL MONOGRAPHS NO. 12

APPENDIX II (continued)

Locality Altitude Sample size Underpart gray number (meters) Mean index Standard deviation

74 1,550 13 6.7 1.55 85 20 10 2.7 0.95 86 320 6 1.0 0.89 87 70 12 2.6 0.79 93 90 6 0.5 0.55 94 300 9 1.9 0.60 95 640 10 2.5 1.08 96 740 10 3.3 1.25 97 1,080 7 3.6 0.69 98 1,330 6** 4.0 2.11 99 380 7 3.1 2.04

* Gray morphs excluded.

** "Brown-napes" excluded. ORNITHOLOGICAL MONOGRAPHS

No. 1. A DistributionalStudy of theBirds of BritishHonduras, by Stephen M. Russell.195 pp., 2 colorplates. 1964. Price $4.50 ($3.60 to AOU members). No. 2. A ComparativeStudy of SomeSocial Communication Patterns in the Pelecaniformes,by GerardFrederick van Tets. 88 pp., 4 text figures.1965. Price$2.00 ($1.60 to AOU members). No. 3. The Birdsof Kentucky,by RobertM. Mengel.Cloth bound, xiv + 581 pp., 4 colorplates plus text figures and vignettes. 1965. Price $10.00 ($8.00 to AOU members). No. 4. Evolutionof SomeArctic Gulls (Larus): an ExperimentalStudy of IsolatingMechanisms, by Neal GriffithSmith. 99 pp., 62 text figures.1966. Price$2.50 ($2.00to AOU members). No. 5. A ComparativeLife-history Study of FourSpecies of Woodpeckers, by Louisede KirilineLawrence. 156 pp., 33 textfigures. 1967. Price $3.75 ($3.00 to AOU members). No. 6. Adaptationsfor Locomotionand Feeding in theAnhinga and the Double-crestedCormorant, by OscarT. Owre.138 pp., 56 textfigures. 1967. Price $3.50 ($2.80 to AOU members). No. 7. A DistributionalSurvey of the Birds of Honduras,by Burt L. Monroe,Jr. 458 pp., 28 text figures,2 colorplates. 1968. Price $9.00 ($7.20 to AOU members). No. 8. AnApproach to theStudy of EcologicalRelationships among Grass- land Birds,by JohnA. Wiens.93 pp., 30 textfigures. 1969. Price $2.50 ($2.00 to AOU members). No.9. MatingSystems, Sexual Dimorphism, and the Role of MaleNorth AmericanPassefine Birds in the NestingCycle, by JaredVerner and MaryF. Willson.76 pp. 1969.Price $2.50 ($2.00 to AOUmembers). No.10. The Behavior of SpottedAntbirds, by Edwin O. Willis, vi + 162pp., 3 colorplates, 27 textfigures. 1972. Price $6.00 ($4.75 to AOU members). No.11. Behavior, Mimetic Songs and SongDialects, and Relationshipsof the ParasiticIndigobirds (Vidua) of Africa, by RobertB. Payne, vi + 333pp., 2 colorplates, 50 textfigures, 40 audiospectrographs.1973. Price $8.00 ($6.40 to AOU members). No. 12. Intra-island Variation in the Mascarene White-eye Zosterops borbonica,by FrankB. Gill, vi + 66 pp., 1 colorplate, 31 textfigures. 1973. Price $2.00 ($1.60 to AOU members). Like all other AOU publications,Ornithological Monographs are shipped prepaid.Make checks payable to "TheAmerican Ornithologists' Union." For the convenienceof thosewho wish to maintaincomplete sets of Ornithological Monographsand to receivenew numbers immediately upon issue, standing orders will be accepted. Orderfrom: Burt L. Monroe,Jr., Treasurer,American Ornithologists' Union, Box 98447,Anchorage, Kentucky 4099•