Ornithological Monographs 12

INTRA-ISLAND

THE

VARIATION

MASCARENE

ZOSTEROPS

BORBONICA

FRANK

ORNITHOLOGICAL

WHITE-EYE

B. GILL

MONOGRAPHS

PUBLISHED

THE AMERICAN

NO. 12

BY

ORNITHOLOGISTS'

UNION

IN


INTRA-ISLAND
THE

VARIATION

MASCARENE

IN

WHITE-EYE

ZOSTEROPS

BORBONICA

BY

FRANK

ORNITHOLOGICAL

B. GILL

MONOGRAPHS

PUBLISHED

THE

AMERICAN

NO.

BY

ORNITHOLOGISTS'

UNION

12


•,A.L

k

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


ORNITHOLOGICAL

MONOGRAPHS

This series,publishedby the American Ornithologists'
Union, has been
established
for major paperstoo long for inclusionin the Union's journal,
The Auk. Publicationhas been made possiblethroughthe generosityof
Mrs. Carll Tucker and the Marcia Brady Tucker Foundation,Inc.
Correspondence
concerningmanuscriptsfor publicationin the series
shouldbe addressedto the incomingEditor, Dr. John W. Hardy, Moore
Laboratoryof Zoology,OccidentalCollege,Los Angeles,California 90041.
Copiesof Ornithological
Monographsmay be orderedfrom the Treasurer
of the AOU, Burt L. Monroe, Jr., Box 23447, Anchorage,Kentucky40223.
(See price list on insideback cover.)
OrnithologicalMonographs,No. 12, vi + 66 pp.
Editor, Robert M. Mengel

SpecialEditor for OrnithologicalMonographsNo. 12, Marion
Anne Jenkinson

Issued May 4, 1973

Price $2.00 prepaid ($1.60 to AOU Members)
Library of CongressCatalogueCard Number 73-78162
Printedby the Allen Press,Inc., Lawrence,Kansas66044


CONTENTS
ACKNOWLEDGMENTS
INTRODUCTION

REUNION

..........................................................................................................

....................................................................................................................

ISLAND .............................................................................................................

vi
1
1

FIELD STUDIES ........ •......................................................................................................

7

BIOLOGY OF ZOSTEROPS BORBONICA ...............................................................................

8

COLOR V•U•L•T•ON IN ZOSr•ROeS

Spec•en exm•ation

sORSOWICA .............................................................

12

............................................................................................. 12

Basis of the color variation ........................................................................
Back coloration .....................................................................................................

12
14

Head colorationin brown morphs......................................................................

21

Underart

30

MtNS•L

coloration ...........................................................................................

V•T•ON

S•c•en

...................................................................................................

39

measurements........................................................................................... 39

Statistics

...........................................................................................................

•alysis

..........................................................................................................

DmCUSS•ON ....................................................................................................................

40

41
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 ...........................................................
Character divergence • natural populations ....................................................

SUMM•Y

........................................................................................................................

L•Ts•

•so

Arrs•g

I

ArrsNDg

II

.........................................................................................................

.......................................................................................................................

...................................................................................................................

51
52
56

59
62

65

ACKNOWLEDGMENTS

This studywas made possibleprimarily throughthe cooperationof the
governmentand citizensof France,Departmentde la R•union. In particular,
I am gratefulto V. Robin,DirecteurdesAffairsG•n•rales,for handlingthe
many administrativeformalitieswith maximum dispatch,and to Armand
Barau, ConseilerG•n•ral de la R•union, Harry Gruchet, Conservateurdu
Museum d'I-IistoireNaturelie de Saint Denis, and ChristianJouaninof the
MuseumNational d'HistoireNaturelie 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 Banquede la R6union,permittedme to
studyand collectwhite-eyeson their properties.The frequentcooperation
of the Servicedes Eaux et For•ts de la R•union, and in particularMessrs.
Moulin, Miguet, and Soroquereis also gratefully acknowledged.Messrs.
Malick and Laidet of the Service M•t•orologique de la R•nnion placed
climatological
data from R6unionat my disposal,and M. ThereseanCadet
generously
providedmostof the plant identifications
mentionedin this study.
In Ann Arbor, Michigan,the membersof my doctoralcommittee,Morris
Foster,Robert W. Storer,HarrisonB. Tordoff, and Warren H. Wagner,providedmany a stimulatingsuggestion.I am indebtedalso to R. B. Bartels
of the Universityof MichiganComputingCenterfor making availablethose
facilitiesandto William Lnnk who paintedthe colorillustration. I have benefitted from discussions

with 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 ScienceFoundationPredoctoral
Fellowshipin 1967 and 1968 and a National ScienceFoundationGrant (GB
3366) to T. H. Hubbellat The Universityof Michiganfor researchin Systematic and EvolutionaryBiology.

vi


INTRODUCTION

Geographicvariationreflectsthe geneticdivergenceof conspecificpopulationsthroughnatural selection;it therebyservesas primary evidenceof the
evolutionaryprocess.Althoughevery local populationdiffers from adjacent
populations
with respectto the frequencyof somecharacteristics,
conspicuous
phenotypicdifferentiationon a microgeographic
scale is evidentprimarily
in terrestrialand relativelysedentaryanimal groups. Geographicvariation
in highly mobile organismssuch as birds, however, tends to involve great
distanceso.relseconspicuous
isolation. The characteristicuniformityof most
localbird populations
seemsto reflectgeneticunitythat is the combinedresult
of gene flow betweenadjacentdemesand the resistanceof coadaptedgene
poolsto fragmentation
by disruptiveselection(Mayr, 1963:296, 361). Yet,
the preciseness

of avian adaptationto climaticgradients(James, 1970) and
the surprisingrapidityof populationcharacterdivergence
in Passerdomesticus
(Johnstonand Selander,1964) continueto emphasizethe overridingimportanceof natural selectionas the primary determinantof patterns of
geographicvariation.
A birdpopulationon a smalloceanicislandischaracteristically
homogeneous
in color and size. Althoughinter-islanddifferentiationis common,intra-island
variation is exceptional. Until the discoveryof four apparent races of a
white-eye,Zosteropsborbonica,on Reunion Island (Storer and Gill, 1966),
Jamaicawas the smallestisland (4,540 squaremiles) known to have geographicallyseparatedbird populationssufficientlydistinct to be considered
subspecies.
ReunionIslandis only about1,000 squaremilesin area. No other
islandwith an area of lessthan 10,000 squaremilesis known to have a polytypic 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
geographic
variation. This studyexaminesthe patternsof phenotypicvariation
in Zosteropsborbonicaon 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 mileseastof Madagascar(Figure 1) at latitude21ø Off S
andlongitude55 ø 30' E, is ellipticalin shape,39 mileslong by 29 mileswide,

and encompasses
about 1,000 squaremiles. It is both the largestof the three
MascareneIslands and the highest,reachingan elevationof 10,068 feet.
1


2

ORNITHOLOGICAL

MONOGRAPHS

NO.

12

Mauritius
-20 $

R•union

-25 $

INDIAN

FretroE

1.

Location of the Mascarene

OCEAN

islands of Reunion and Mauritius

in the western

Indian Ocean. The third Mascarene island, Rodriguez, is located 250 miles to the
east.

Reunion'sfaunalrelationships
are with Madagascarand the otherislandsin
the westernIndian Ocean. The islandwas colonizedprimarilyfrom Madagascar(Moreau, 1957:399) although.,judgingfrom the similarityof their
faunas,irtterchange
betweenMauriflusandReunionislandsmusthaveoccurred
quitefrequently.At least8 speciesof birdsbecameextinotshortlyafter man's
arrivalon Reunionin 1663. Todaythe land bird avifaunaincludes11 indigenous
and 16 introducedspecies.Discussions
of thesespeciesare available
in Milon (1951), Berlioz (1946), and Watsonet al. (1963).
Topographically
ReunionIslandis an extremelyruggedisland,risingsteeply


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBONICA

3

ALTITUDE
PIT
GRA

ST.

N D

0 N

NEIGES

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.astal

plains(Figures2 and 3). It is composedof two overlappingshieldvolcanos(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-headed
valleyerosion"(Upton and
Wadsworth,1966:5), is dissected
up to depthsof 2,500 metersby deepgorges
and threehugebasin-like"cirques." The youngestlavaseruptedfrom Piton
des Neigesbetween100,000 and 350,000 years ago in the late Pleistocene
(Chamalaunand McDougall, 1966).
The locationof ReunionIslandat the southernedgeof the tropicsensures
a rather stable,mild climatewhichis dividedinto a warm rainy seasonfrom
Novemberto April, and a cool dry seasonfrom May to October. Midsummerdaysaveragetwo and one-halfhourslongerthan winter days and
five to six degrees
centigrade
warmer. Averagemonthlyrainfallon the windward sideof the islandis about 100 cm higherin the wettestmonth, March,
than in the driestmonth,August. The differenceon the island'sdry sideis
about

15-20

cm.

ReunionIsland'sruggedrelief promotesmarkedlocal variationin 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 easternwindward slopesit reaches800 cm a year. Temperaturedecreasesabout 7øC
with each1,000 metersaltitude,or about2øC per 1,000 feet. Highlandsover
2,000 metersare 14ø to 16ø colderthanthe coast,wheretemperatures
average

about 22øC in the summer.

Rainfall data and temperaturerecordsin the form of ten-yearsummaries
for 112 and 30 Reunionlocalities,respectively,
were made availableby the


ORNITHOLOGICAL

4

MONOGRAPHS

NO.

12

ST. DENIS

I
Pos

?eBr•l•
:hicots

ST. PAUL

ST.BENOIT

• CIRQUE

Ja

Petite

•

France

•

DE

SALAZIE

Piton
•

des Neig,•

ROSE

ForGt
du

BGbour

For•t

DE

Mourouvin

Piaine
des
Cartes

ST.
TSvelave

Blanc
LE

Etarig Sal•_•_
les

Bains

•,•

ST. LOUIS

liv. St.
Etienne

ST. PI

,o

s,

i?

kilometers

PHI L I P PE

ST. JOSEPH
FIOURE 3. Reunion

Island

localities.

ServiceM6t6orologiquede la R6union. For a few additional localities for
whichrecordshave beenkept only a few years,I computedaveragesfrom the
annual "BulletinsClimatologiques"pubh'shedby the same service. Many of
my collectinglocalitieswere sufficientlyfar from an actualmeteorological
station that I usedrainfall valuesfrom isohyeticalchartsbasedon the ten-year
summaries
andcalculatedtemperatures
regressed
from existingaltitudinalcorrelations.Two temperatures
have beenusedin this studyfor each locality:
the mean maximum of the warmest month of the year (either January or
February) and the meanminimumof the coldestmonth of the year (usually
August). Altitudeswerefixed to the nearest20 meterseitherfrom existing
altitude markers or from a 1:50,000 topographicalmap publishedby the
MinistbredesTravaux Publicset desTransports(Institut GeographiqueNational) of France.

Corresponding
to variationin localclimatesis a diversityof habitats.The
indigenousvegetationis describedby Rivals (1952) and its distributionon
the islandbefore the arrival of man is illustratedin Figures2 and 5. The


1973

•

GILL:

IN

ZOSTEROPS

BORBONICA

5

< •oc

<
•

VARIATION

/

<•00

<400

•

'<500

•

>soo

FIeURE 4.

Distribution of annual rainfall on Reunion Island, Isohyets are in cm.

vegetationof certainstudyareasis describedin Gill (1971). Originally there
was a tree savannain the arid northwesternlowlands (below 400 meters elevation), whilea tamarin(Acaciaheterophylla)forestrich in epiphytesandferns
occurredin protectedlocalitiesbetween 1,200 and 2,000 meterselevation,
and high altitude heath was presentabove 2,000 meterselevation. Most of
the island, however,was coveredby complexevergreenforest with trees
about10 metershighandrarelyexceeding15 meters. It includedmany species
with irregulardistributions,and could be subdividedon the basisof predominantspeciesand prevailingclimate.
The indigenousvegetationhas been almost completelydestroyedbelow
900 meters elevation on the w•st side of the island and below 400 meters

elevationon theeastside;it hasbeenreplacedby sugarcanefields,pasturage,
andtheexoticplantsthattypicallysurroundhumanhabitations.Scrubvegetation or thicketsof variousmixturesof heath (Philippia), guava (Psidium),


6

ORNITHOLOGICAL

FroORE5.

MONOGRAPHS

NO.

12

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 representsrecentlyformed lava which usually lacks vegetation.

Lantana, and Rubushave replacedthe originalforestsin many cutoverareas.
Althoughthe proportionsof the componentspeciesmay vary greatly,this

vegetation
tendsto be distinctfrom otherhabitatsin that it is relativelyopen
with no overheadcanopyand is low and densein over-allstructure.Physiognomically
it is closest
to thelow ericaceous
vegetation
of thehigh altitudes
andoftenincludes
someof thesameplantspecies.Thelowlandravinescontain
lushgrowthsof Eugenia]ambos(Myrtaceae)on the eastsideof the island

and of $chinusterebenthifolius
(Anacardiaceae)and Pithecollobiumdulce
(Leguminos.ae)
on thedry westside. On the westcoastnearEtarigSal• les
Bainsa largetract of forestwith Casuarinaequisetifolia
(Casuarinac•ae),
Pithecollobiumdulce,Melia azedarach(Meliaceae), and $chinusterebenthi[oliusas the dominantplants,replacesthe otherwisebarrencanefieldsand
goatpasturage
of the area. Alongthesoutheastern
coastfromSte.Roseto St.
Josephrelativelyundisturbedforest composedprimarily of Mimusopsspp.


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBON1CA

7

(Sapotaceae)descendsto about 200 meterselevation. The recentlava flows
and young,poor soilsof this area have not encouragedagriculturalexpansion.
The forestsof Grand Brill6 on the southernslopesof le Volcan representa

pioneerstageof colonizationof the new lava and are composedprimarily of
Casuarinawith a fern (Acrostichum)understory.
Much of the indigenous
vegetationat higheraltitudesis still intact although
one is never far from somehuman disturbanceor exotic species.Between
900 metersand 1,400 meterselevationon the westernslopesof Reunionthe
cultivationof geranium(Pelargonium),for its essentialoils, has led to widespreaddamageto theforest,whichwasclearedin plotsfor plantingand which
providesfirewoodfor thedistillationprocess.In someareasat moderateelevations, e.g., la Petite France (1,100 meters) and Plaine des Cafres (1,200
meters), the introducedwhite acacia(Leucaenaglauca) is now predominant.
The wet centralforestsof B6bourare relativelyundisturbed
althougha dense
undergrowthof in.troducedFuchsia coccinia is establishedin much of this
area. Finally the steep, relatively inaccessiblecliffs throughout the island,
includingthe sidesof the otherwiseheavilypopulatedcirquebasins,still retain
muchof theiroriginalvegetation.The highaltitudeheathvegetation
is perhaps
the leastchangedof Reunion'soriginalhabitats,althoughsome,especially
near

PlainedesCafres,hasbeenconvertedinto 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 originalcognizanceof
the problem. Field work in 1967 includedintensivesamplingthroughout
the islandand studiesof the white-eye'shabits,throughobservations
of colorbandedindividuals. I alsomaintainedcaptiveZ. borbonicaon Reunion Island

andat theUniversityof 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 localitynamesand coordinates
were taken from the topographical
map mentionedearlier.
The few specimens
of Zosteropsborbonicacontainedin major collections
are labelledonly "ReunionIsland" and consequently
had little or no value
for the presentstudyof variationwithin the island'sconfines.As a resultall
specimensused in this study were obtainedduring the two above-mentioned
expeditions.The 1964 specimens
have beendepositedin the U.S. National
Museumcollections
and the restin the collections
of The Universityof Michigan Museum of Zoology. In all, 759 specimensfrom 76 differentlocalities
were taken. I would remind thosepersonswho may feel someconcernabout
taking specimensof an endemicbird speciesfrom a small oceanicisland not


8

ORNITHOLOGICAL

MONOGRAPHS

NO.

12

11

10

1

2

12

57
3

•17

^

97 72

66

16

•18

15• ?.4

25
99

, 20 21

7470

19

23

268090 82•

56

101

64
63

IO0

42

62

98

52

95 61
94

51

54

0 5 I0
KILOMETERS

FIGURE 6.

41
14
3

5••.4,•.
91 •
34

86 •

Localities visited on Reunion Island (see Appendix I).

only that Zosteropsborbonicais abundanton ReunionIsland, havingundoubtedlybenefitedfrom humandisturbancerather than suffered,but also
thaXthe speciescontinuesto flourishdespiteannual predationby local inhabitants.

Exceptfor the monthof September,specimens
were taken continuallyfrom
May throughDecember.Many of thosetakenduringthe breedingseasonfrom
Octoberto Decemberare badly worn and have consequently
been excluded

from mostanalyses,aswereany otherspecimens
in similarcondition.
BIOLOGY

OF

ZOSTEROPS

BORBONICA

Zosteropsborbonicais an endemicMascarenespeciesfound only on the
two main islands,Mauritius and Reunion. Althougha typical white-eyein
proportionsand behaviorit is somewhataberrantin plumagecoloration,
having lost both the familiar white eye-ring and the carotenoiddeposition
responsible
for the characteristic
greensand yellowsof Zosteropids.Furthermore,the species
hasacquiredwhite "anklets,"a distinctivewhite rump patch
formedby the white upper tail coverts,and white axillary featherswhich are


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBONICA

9

frequentlyexposedlike sunbirdpectoraltufts. Zosteropsborbonicawasonce
includedin the genusMaIaciropswith the gray white-eye(Z. modesto)of
the SeychellesIslands, but Moreau (1957:400) maintainedthat the few
plumagepeculiarities
distinguishing
thesespecies
did not merit separaterecognition at the genericlevel. He consequently
synonymized
Malaciropswith
Zosterops (see also Moreau, 1967). To date there is no evidenceas to the
affinitiesof Zosteropsborbonica.

Color variationin ReunionZosteropsborbonicawas recognizedin various
early works,but its naturewas subsequently
the sourceof muchl•axonomic
confusionand speculation.Pollenand van Dam (1868:74), the first authors
to recognizethe extentof the color variation,thoughtit resultedfrom individualbut not sexualdifferences.Shortlythereafter,Hartlaub ( 1877:95)
restricted
the name"borbonica"
to brown-backed,
gray-headed
formsand,on
the basisof two specimens
in the Newtoncollections
now at CambridgeUniversity,described(pp. 97-98) a secondspecies,ZosteropsE. Newtoni, of

whichthe presumedmale was all gray and •thepresumedfemale rich brown
on boththebreastandunderparts.Gadow(1884:195-196) did not recognize
ZosteropsE. Newtoni on the groundsthat there was a perfectgradationbetween"borbonica"
andthisformin thesmallseriesof specimens
he examined,
but he considered
the all-grayform to be the male,the brownform to be the
female,andthegray-headed,
brown-backed
formto be theyoungof Zosterops
borbonica,corresponding
to male Z. E. Newtoni, female Z. E. Newtoni,
and Z. borbonicaof Hartlaub, respectively.Gadow also suggested
that the
brown-backed
form couldrepresent
the winterplumageof the species
rather
than the adult female. Newton and Newton (1888) criticizedthese conclusions
andsupported
Hartlaub'searlierrecognition
of two species.
Patternsof geographicdistributionof the variouscolor forms were first
described
by Storerand Gill (1966), who delineatedfour subspecies
(see
frontispiece)from ReunionIsland: Z. b. borbonica,a gray-headed,
brownbackedformof theisland'snorthernlowlands;
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
underparts
fromthecentralforests;
andZ. b. edwardnewtoni,
an all grayform
from the highlandheathzone. Moreau (1967:333) maintainedthe conventional pair of subspecies,
Z. b. mauritianaon MauritiusIsland and Z. b.
borbonica on Reunion Island.

Unlike most remainingendemicReunionbirds, which dependon the
remnantsof indigenousforest, Z. borbonicais common in most habitats,

includinggardens,evergreen
forests,Casuarinaforest,and highlandheath.
But despite
its seemingly
broadhabitattolerances,
it is primarilya species
of
openand disturbedareasratherthan intactforests;its presentdistribution
reflectsthe widespreaddestructionof original habitats. In tracts of un-


10

ORNITHOLOGICAL

MONOGRAPHS

NO.

12

disturbedinteriorforestsit is decidedlyscarceexceptat the edgesof large
clearingsand along streambeds.
Like mostwhite-eyes,Z. borbonicais a socialspeciesusuallyencountered
in partiesof 6 to 20 individuals. Suchflocks are formed throughoutthe year
a}thoughthereis a tendencyfor the birdsto breakup into pairsat the heightof
the breedingseason.I rarely observedintraspecificaggression;
it neverseems
to be associatedwith territorial or nest site defense. When groupsof feeding
Z. borbonicapasscloseto activeneststhey are joinedby the nestingbirds,not
chased,and nestsare sometimesbuilt within 50 feet of each other in a loose
colonial arrangement.
Throughoutthe year individualZosteropsborbonicafrequentlyhuddle
closelytogetherin pairsor groupswhilerestingor engaging
in mutualpreening.
Suchallopreeninghas been notedin other speciesof white-eyes(Harrison,
1965:195; Eddinger,1967). I have neverwitnessed
any aggression
in Z.
borbonicathat was the resu}tof proximity. Rather, a white-eyeavoidsthe
huddlingactionsof anotherindividualsimply by leaving. Only rarely does
a singlerestingwhite-eyerefusean opportunityto huddle,andin the Mascarene
forestsit is commonplace
to seetwo white-eyessidlealonga branchuntil they
touch each other.

Z. borbonicafeeds primarily on insects,but eats a variety of fruits
particularlyduringthe winter. Flower feedingis frequentat certainfavored
flowers. A detaileddiscussion
of borbonica's
feedingecologyis availablein
Gin (1971).
Zosteropsborbonicacoexistson Reunion Island with' Zosterops o,livacea,
whichis an aggressive,
nonsocial,
specialized
flowerfeeder. Zosteropsolivacea
feedsprimarilyat flowers,catches
insectsby fly-catching,andrarelyeatsfruit.
It is usuallyfound in pairs, not flocks,and defendsselectedfloweringplants
againstbothZ. borbonicaand otherZ. olivacea.Zosteropsolivaceaalsois
more closelytied to the complexforestsof the centralhighlandswherethere
is a varietyof suitableflowers. Zosteropsolivaceaseemsto have beenthe
first white-eyeto colonizethe MascareneIslandsand Zosteropsborbonica
arrived sometimelater. The differencesbetweenthe .two speciesand their
probableevolutionare discussed
in detailelsewhere(Gill, 1971).
Breedingin Zosteropsborbonicais seasonalas it is in most Mascarene
species
and occursduringthe australsummer.In earlyAugust,1967, I noteda
suddenincreasein testicularsizein thesebirds; enlargementcontinueduntil
full size,averagingabout7 x 5 mm, wasreachedin earlyOctober. Preceding
this morphological
indicationof breedingreadiness,however,were noticeable
changes
in behavior.Startingin the lastweekof Julyintraspecific

chaseswith
repeateddisplacement
were regularin feedingwhite-eyeflocks. Cer[ain apparentlysubordinate
individualsoften quiveredtheir wingseven when not
beingchased,
andaggressive
individuals
oftenprominently
displayed
theirwhite


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBON1CA

11

axillary tufts and rump patch. Pronouncedvocal activityincludingelaborate
singingbeganat aboutthe sametime as•theintrasp,ecific
chasing.
Nestingbeginsin late SeptemberandOctoberandcontinuesthroughDecember, perhapsinto J.anuaryand February. In 1967 I noted the first fledged

youngon 14 Octoberand foundboth fledgedyoungand evidenceof nestlings
up to thelastday of field work on 21 December.The peak of breedingactivity
seemsto occurin October and November; local residentsconfirm this observa-

tion. While I haveno evidenceof doublebroodsthe presenceof manynests
with eggsin December,1948 (Milon, 1951:141), suggests
that somepairs
may breedmorethan oncea year.
Althoughthe generaltimingof breedingwas the samethroughoutReunion
Island from SeptemberthroughDecember,there were some indicationsof
local variability. I found nestswith eggsin the secondweek of Septemberin
thewetlowlandsat Beaufonds(Ste.Marie). Yet in the highlandsat le T•velave
( 1,700 m) and Cilaos (900 m) I did not observeactivenestbuildinguntil the
first week of Octoberdespiteregular checks. On the dry west coast there
was.anabortivenestingattemptat EtangSa16lesBainsin mid-September
that
coincided
with .anabnormaland temporaryburstof growthof the vegetation,
but elsewhereon this coast (e.g., St. Paul) nest buildingwas not observed
until the end of October. Thusit seemslikely that localpopulations
breedin
accordwith local seasonality,
whichmay staggerthe onsetof nestingas much
as a month between different localities.

Cooperativefeeding of the young takes place at least in certain circumstances.The only Z. borbonicanest I watchedfor an extendedperiod was
onein the forestsat Cilaoswith two youngthat were almostreadyto fledge.
At least four adultsbroughtfood to theseyoung and, of these,three were
collectedand proved to be males with ossifiedskulls and enlargedtestes.
Followingfledgingof theyoung,family groupsoftenjoin, at leasttemporarily,

the feedingflocksin their area. Isolatedfamily groupsoccasionally
included
an extra male that activelyfed the young. Similar casesof cooperafivefeeding
are knownin a varietyof birds,particularlysedentaryspeciesin the tropics,
andusuallyinvolvecloselyrelatedadults(siblings)or youngof the previous
year (Lack, 1968:80).
In Zosteropsborbonicaa complete,prebasicmolt follows breeding. Of
83 adult borbonicaspecimens
taken in December,13 were molting body
feathers. PresumablyJanuary through March, the warmestmonthsof the
year, are the monthswhenthe majorityof the populationmolts,but I have
no informationfor thisperiod. Of the 143 spedmenshandledin May, 15 were
completing
their molt with tracesof body molt and moltingfeathersin the
crownbeingevident. All borbonicacollectedor bandedat this seasonwere in
fresh,unwornplumage.No moltof anykind wasevidentduringJuneandmost
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 borbonicabandedin August and

September.This molt resultsin no changeof the plumagecoloration. I have
no informationon the extent of the postjuvenal(first prebasic) molt which
presumablytakesplacefrom Januaryto March at the sametime as the adult
prebasic molt.
Except for local concentrations
of individualsat abundantfood sources,
populationsof Z. borbonicaappearto be sedentarythroughoutthe year. First
therewere no major changesin the distributionof white-eyesduringmy field
work from April to December,1967. Also, individualscolor-markedduring
the winter monthsat severallocalitieswere resightedat thesesame localities
duringthe 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 samehollowduringtwo hoursof observationin early September.Finally
11 of 37 individualsbandednear Cilaos,in late July and early August,were
seenagainbetween16 Octoberand 20 Decemberin the sametract of forest.
These observationsindicate that while Zosteropsborbonicamay 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,showsno aptitude for catching small forest birds, but feeds
insteadon introducedmammalsand gallinaceousbirds. In the lowlands,nest
predationby rats, an introducedsnake (Lycodon aulis), and an introduced
lizard (Calotesversicolor)hasbeenreportedby local residents.Althoughthe
practiceis officiallydiscouraged,
largenumbersof white-eyesare still caught
for foodby the nativesof ReunionIslandwith bird lime.
COLOR

VARIATION

IN

SPECIMEN

ZOSTEROPS

BORBONICA

EXAMINATION

I evaluatedcolor charactersvisually againsta standardgray background
in northsky1/ghton brightto cloudybrightgaysin the morningin Ann Arbor,
Michigan. I evaluatedeach characterindependentlyof the others and subdividedi,t into a seriesof characterstatesthat includedas many intermediate
categoriesas I could consistentlydistinguish(Table 1). I comparedeach
specimento a referencespecimenwhich representedeach character state.
I alsostudied,throughmicroscopic
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


GILL:

1973

VARIATION

IN

ZOSTEROPS

TABLE

13

BORBON1C.4

1

INDEX TO COLOR CHARACTERS
IN Zosterops borbonica
Character

Back

Character

Reference
specimenx

213 784
0 - pure gray
213 524
1 -gray with slight scatteredbrown
2 - heavy suffusionof brown in gray--nearly 213 469

color

a 50:50

Head

states

mixture

3- brown with scattered gray feathers
4 - entirely brown

213 410
213 779

0 - entirely gray
1 - gray with scattered brown
2 - 50:50 gray-brown
3 - brown with gray wash
4- strong, pure brown
0 - pure gray
1 - gray with scatteredbrown
2 - 50:50 gray-brown
3 - brown with gray wash
4 '- strong, pure brown
0 - entirely gray
1- ocular gray and postauricular gray continuous and broad with only slight brown
2- well-defined gray above auricularsto eye

as well as patch of gray on sidesof neck,
though separatedfrom ocular gray by a

213
213
213
213
213
213

color

Crown-forecrown

Nape

Face

brown

Auriculars

Underpart color
Amount of gray

213
213
213
213

493
330
609
514
608

213
213

577
569

213 536

collar

3 - diffuse gray around auriculars only
4- entirely brown, no discernible gray
0 - pure gray
1 -slight brown, especially below eye
2 - 50:50 brown-gray
3 -predominately brown but with some gray
4 - entirely brown

213 874
213 661

213
213
213

213
213

567
470
615
727
392

0- white

213
213
213
213
213
213
213
213
213

901
512
352
427
309
331
415
384
419

213
213
213
213
213

496
785
411
782
341

1-

2-

3-

increasing

4-

qntensities

56-

of gray

78-

Amount

500
601
599
826
343

of brown in breast 01-

234-

dark gray
absent
faint brown wash

strong but narrow band
strong wide band
strong and extensive, including parts of
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.

onthepresence
andamountof twotypesof melanir•s:eumelanins,
whichare
blackishgranules,
and phaeomelanins,
whichare reddishor light brown
granules.The eumelanins
of Zosterops
borbonica
are typeB granules(see
Moreau,1957:323) deposited
in denseconcentrations
in the barbulesandto
a lesserextentthe rachisof eachfeather,but not the barbs(Figure 7). The

phaeomelanins
correspond
to Moreau'stypeC granules
andaredeposited
in
the barbs and rachis but not the barbules. The basic difference between gray

andbrownin the plumageof Zosterops
borbonica

liessimplyin the addition
of phaeomelanins
to a featherwitheumelanin
to producebrown.
Severalimportantaspectsof the color variationdependon where
phaeomelanins
aredeposited
in theplumage.Other,essentially
independent,
aspects
of thecolorvariation,
particularly
the colorof the underparts,
result
fromvaryingconcentrations
of bothmelanins.Thereis no variationin color
dueto sexor agedifferences
oncethedefinitive
plumage
is attained.
BACK COLORATION

Dimorphism.-•I distinguished
five categoriesof back colorationin my

specimens
of Zosterops
borbonica
from ReunionIsland(Table 1). Of the
343 adultspecimens

fromlocalitieswith morethan onephenotype
for back
color,301 (87.7 per cent) areeitherpuregray-backed
(Category0) or pure
brown-backed
(Category4), whereasonly42 specimens
(12.3 per cent) are
intermediate
(Figure8). Of the42 intermediate
specimens,
30 wereclassified
in Category1, 5 wereput in Category3, and only 7 (2 per cent) were


1973

GILL: VARIATION IN ZOSTEROPSBORBON1CA

15

8O

6O

4O

2O

o

1

BACK

2

3

4

COLORATION

FIormE8. Backcoloration
of 343 adultsfrom localities
knownto havemorethan

onebackcolorphenotype.
Category
0, "pure"graymorphs;
Categories
1-3,
intermediates;
Category
4, "pure"brownmorphs(seeTable1).

markedly
intermediate,
i.e.,in Category
2. Thusthischaracter
is strongly

dimorphic,
withonlyaslight
tendency
towards
intermediacy,
andintermediacy
usually
involves
scattered
traces
of brown
pigmentation
in otherwise
graybacked
individuals.
Puregray-backed
individuals
tendtohavetraces
ofbrown
elsewhere
intheplumage
particularly
ontheflanks,
andspecimens
in intermediate
backcolorcategories
.show
corresponding
increases
in theintensity

of brownin theseareas.Unless
otherwise
stated,
whenI usetheterms"gray

morphs"
and"brown
morphs"
I amreferring
toindividuals
inCategories
0-1
and 3-4, respectively.

Geographic
distribution
of grayandbrownmorphs.--The
proportion
of
graymorphs
wasascertained
forthe83 localities
forwhich6 or moreindividuals
(specimens
andsightrecords
combined)
wereavailable.
The 45 localities
at whichgraymorphsor intermediates
wererecorded

comprise
a large,
continuous
region
in thecenter
of theisland
(Figure
9);
noneare at coastallocalities.At onlyonelocality(number42 at 2,300

meters)
didI findonlygraymorphs;
nospecimens
weretakenbutall nine
birdsobserved
therewere assignable
to categories
0 or 1. In contrast,many
lowlandlocalitieshaveonlybrownmorphs.

I counted
thenumbers
ofgrayandbrownmorphs
at several
localities
during

theearlywinterandagainat thebeginning
of breeding

season
(Table2).
Somevariationbetween
counts,dueprimarilyto sampling
error,is present


16

ORNITHOLOGICAL

MONOGRAPHS

NO.

12

o
o

0

19

9

0

0

0

5O

2

•9

0 2125
626,

18

"•27
200

7

.0
0

12

.0

½ i0
16

27
9O

47

31

02

0

5

0TM

61

I0

KILOMETERS

0

0

Fioum• 9. The percentageof 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 consistency
of the resultsfor each
localityis evidenceof seasonalstabilityof the ratio of the two morphs.
An exceptionto thisgeneralobservation

wasnotedat Nez de Boeuf (Plaine
desCafres) at 2,060 meterselevationon the edgeof the Ravine de la Rivigre
des Reinparts, where an endemic leguminoustree, Sophora nitida, whose
flowersseemespeciallyattractiveto white-eyes,was abundant. The relative
frequencyof gray morphsat this localityincreasedfrom 30 per cent to about
70 per cent betweenAugustand early September,1967, as the $ophora
flowerswaned (Figure 10). The stabilizationof the morph ratio after early
September,as well as my observations
of white-eyesmoving up and down
the adjacentcliffs during the day at the height of the Sophoraflowering,
indicatethat brownmorphsmoveup temporarilyfrom the adjacentlowlands
to feed at these flowers.

Altitude is the mostimportantfactor affectingthe distributionand relative
abundanceof gray morphs. A regular altitudinalincreasein the percentage
of graymorphsoccurredalongfive transects
I madefrom the coastto the high-


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBON1CA

17

landsin areasof relativelycontinuous,
undissected
topography(Figures11, 12,
and 13). On four of the five transectsthe relative abundanceof gray morphs
increasedcontinuouslywith altitude above 1,350 meters. On the fifth
transect(St. Benoitto la PlainedesPalmistes--opentrianglesin Figure 13),
the relative abundanceof gray morphsincreasedcontinuouslyabove 720
meterselevation;the predominance
of openscrub,whichis physiognomically
similarto heath,ratherthanforestat low altitudesmay have beeninfluential
becausescrubvegetationwaspresentat all non-cirquelocalitiesbelow 1,300
meterselevationwheregraymorphswererecorded(Table 3). Thisis evidence

of specifichabitatpreferences
by gray morphs. With only one exception
(locality52) high gray morphfrequencies(over 60 per cent) were not recorded below 1,600 meters elevation (Table 3).

Topography
alsoaffectstheproportions
of graymorphs.This canbe seen
in samples
from insidethe deeplyerodedcirques.The data availablefor the
Cilaos and Salaziecirques(Figure 14) showhigh gray morph frequencies
at relativelylow elevationsand markedfluctuationsin the percentageof gray
morphsat differentaltitudes.A characteristic
featureof cirquetopographyis
the separationof highaltitudes,wheregraymorphsarecommon,from very low

altitudes,
wheretheyarenormallyabsent,by sheercliffs. Thisgreatreduction
of distances
separatingextremealtitudespromotesunpredictable
mixturesof
TABLE

2

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

Date

Number
of gray
morphs

4

4-6 May
9 September

5

7 June

24

9 September

44

19 April

2

20 June
3 November

27

12 May
26 November

40

16

58

73

10
10

Number
of brown
morphs

7
10

Per cent
gray

Type of record

58.8
50.0

Banded birds
Sight

5

82.7

Sight

16

73.3

Sight

10

16.7

Sight

2
2

10
8

16.7
20.0

Specimens
Sight

0
0

21
12

0
0

28 April-1 May

15

29

34.1

Banded birds

6 September

10

14

41.6

Sight
Sight

Sight and specimens
Sight and specimens

15 June

33

53

38.4

26-29 July

11

17

39.3

Banded birds

17 October

10

15

40.0

Sight

9

1

90.0

Specimens

12

1

92.3

Sight and specimens

25 May
5 November '


18

ORNITHOLOGICAL

MONOGRAPHS

NO.

12

•oo,

•,• •o,
b.i

o

0>6 70
•
60
On,'
U.ILu

E_• 50

o,•

:2a:

30

•[

zo

o

IO

I

I

I

'•

7

17

5

23

AUG

AUG

SEP

OCT

I ,

I

22

18

NOV

DEC

FmURE10. Temporalchangein the ratioof graymorphs(dashedline) andabundance
of $ophoraflowers(solid line) at Nez de Boeuf. The data are for 1967, except
those for November are from 1964.

colorforms,becauseanymovementup or downthe cliffsby white-eyes
would
quicklyalterthe situationexistingasa resultof altitudinaleffects.
Behavioralinteractions
o/gray and brownmorphs.---Ihaveanalyzedseveral
behavioralfeaturesto dexerminewhetherbrown and gray morphsassociate

non-randomly.On 9 September
on the PlalnedesChicots,whenwhite-eyes
werecominginto breedingconditionbut prior to actualnesting,I encountered
many isolatedforagingpairs (bisexual?)of white-eyesand recordedthe
morphsinvolved.At CitaosI recorded
themorphsinvolvedin all intraspecific
chases
notedearlyin the breedingseason,as well as for established
pairsof
breedingadults,andpairsof white-eyes
engaged
in huddlingand allopreening.
The expectedfrequencies
of variouspossibleassociations
werecalculatedfrom
theobserved
percentages
of bothmorphspresentin .thetwoareas.The observed
andexpectedfrequencies
werethentestedfor significance
by useof a x2 test
(Table 4). Boththe associated
pairsof white-eyes
at PlainedesChicotsand
the matedpairsat Cilaosprovedto be randomwith respectto plumagetype.
On the other hand huddlingsand chaseswere recordedmore frequentlybetweenlike morphsthan was to be expectedif the partnershad been chosen
randomlyfrom the total local population. Interpretationof theseresultsis
clearlydifficult;howevera lackof morphdiscrimination
in relationto mating,
asis suggested

by the randompairingassociations
at Cilaos,would resultin
the maintenanceof polymorphism.
Geneticso/back coloration.--JuvenalZosteropsborbonicahave patternsof


1973

GILL:

VARIATION

IN

ZOSTEROPS

BORBON1CA

19

upperpartcolorationsimilarto thoseof the adults. Someare entirelygray corresponding
to the pure gray morphadultsand have similarvariationsin the
presenceof brown in other parts of the plumage. Others are brown and
equivalentto brownmorphadultsin bothback and headcoloration.
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
establishing

that the adultspresentwere feedingthe young, I collectedthe

ß
ß
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 Bainsto Piton desEpinardsvia le T•velave (Localities51, 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
Salazesvia 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).