COMPETITION
NATIVE
BETWEEN
EUROPEAN
WOODPECKERS
STARLINGS
FOR NEST
AND
CAVITIES
IN SAGUAROS
THEODORE A. KERPEZ 1 AND NORMAN
S. SMITH
U.S.Fishand WildlifeService,ArizonaCooperative
Fishand WildlifeResearch
Unit,
Universityof Arizona,Tucson,
Arizona85721USA
ABSTRACT.--European
Starlings(Sturnus
vulgaris)have recentlyinvadedArizona and breed
in someareasbut not in similarareasnearby.In Arizona,EuropeanStarlingscommonlynest
in cavitiesin saguarocacti(Carnegiea
gigantea)
but do not excavatethesecavities.To examine
whetherEuropeanStarlingscompetewith nativewoodpeckers
for nestcavitiesin saguaros,
we studiedGila Woodpeckers(Melanerpes
uropygialis)
and Northern Flickers(Colaptes
auratus)
in areaswith EuropeanStarlingsand in similar, nearby areaswith no EuropeanStarlings.
We determinedwhich factorsexplainedthe variationin the numberof nestsof eachspecies
presenton fifteen 10-haplots.We alsocomparedthe locationand dimensionsof nestcavities
usedby eachspeciesto determinewhether EuropeanStarlingsuse Gila Woodpeckernest
cavities, Northern Flicker nest cavities, or both.
We foundthat EuropeanStarlingscompetewith Gila Woodpeckers
but not with Northern
Flickers.Thiscompetitiondecreases
the numberof Gila Woodpeckers
that nestin areaswhere
EuropeanStarlingsnest.EuropeanStarlingsusedGila Woodpeckernestcavities,and there
wasa negativerelationshipbetweenthe numberof EuropeanStarlingnestsandthe number
of Gila woodpeckerneststhat explained46.7%of the variationin the numberof Gila Woodpeckernestson the plots.EuropeanStarlingsdid not useNorthern Flickernestcavities,and
we found no relationshipbetween the number of EuropeanStarling nestsand the number
of Northern
Flicker
nests.
In addition, the number of Gila Woodpeckernestswas positivelyrelated to the number
of large saguarosand negativelyrelated to the slopeof the plot. The number of Northern
Flicker nestswas positively related to the volume of ironwood (Olneyatesota).The number
of EuropeanStarling nestswas negativelyrelated to the distanceto agricultureand large
lawns. Received29 June1989, accepted4 December1989.
EUROPEAN
Starlings(Sturnusvulgaris)were introduced into North America in 1890 and rapidly spreadthroughoutmostof the United States
(Kessel 1953). They were not observed in Arizona until ca. •946 (Monson 1948). In Arizona,
Woodpeckers.One pair of Gila Woodpeckers
lost three successivecavities to European Star-
lings. EuropeanStarlings have been observed
usurping nest cavitiesfrom Northern Flickers
in New Hampshire (Shelly 1935), Maryland
EuropeanStarlings commonly nest in cavities (Howell 1943), and Massachusetts(Bent 1950).
in saguarocacti(Carnegiea
gigantea),but they do EuropeanStarlings have also usurped nest cavnot excavate these cavities (Kessel 1957). Gila
ities from many other speciesincluding Purple
Woodpeckers (Melanerpes uropygialis) and Martins (Prognesubis)in Michigan (Allen and
Northern Flickers (Colaptesauratus)excavate Nice 1952), Red-bellied Woodpeckers(Melathese cavities in saguarosfor nest sites(Gilman nerpescarolinus;
Kilham 1958) and Wood Ducks
1915, Bent 1939), and the nesting seasonof the (Aix sponsa;
McGilvrey and Uhler 1971) in MaryEuropean Starling overlaps that of the Gila land, Acorn Woodpeckers(MelanerpesformiciWoodpeckerand the Northern Flicker (Gilman vorus)in California (Troetschler 1976), Eurasian
Nuthatches (Sitta europaea)in Sweden (Nilsson
1915, Bent •939, Royall •966).
Brenowitz (1978) observed European Star- 1984), and Buffieheads (Bucephalaalbeola) in
lings usurpnestcavitiesfrom three pairsof Gila British Columbia (Peterson and Gauthier 1985).
Van Balen et al. (1982) concluded that by competing for nest cavities European Starlings de•Present address:Department of Fisheries and creasedthe number of Great Tits (Parusmajor)
Wildlife Sciences,Virginia PolytechnicInstituteand nesting in their study area in the Netherlands.
StateUniversity, Blacksburg,Virginia 24061 USA.
Our objectivewas to determine whether Eu367
The Auk 107: 367-375. April 1990
368
KERPEZ
ANDSMITH
ropean Starlings competed with Gila Woodpeckersand Northern Flickersfor next cavities
in saguarocacti,and if so,to evaluatethe effects
of this competition. European Starlings pres-
ently breedin someareasof Arizona but not in
similar areas nearby. Therefore, we were able
to studyGila Woodpeckers
and Northern Flickers in areas with no European Starlings and in
similar, nearby areaswith various densitiesof
EuropeanStarlings.To assessthe effectof competition, we studied the relationship between
the number of EuropeanStarlingsnestingin an
area and the number of Gila Woodpeckersand
Northern Flickers nesting in that area. We examined the relationshipsbetween habitat variables and the numbers of nesting Gila Woodpeckers, Northern Flickers, and European
Starlings to separatethe effectsof habitat and
competition.We alsocomparedthe locationand
dimensionsof nestcavitiesusedby eachspecies
to determine whether European Starlings use
Gila Woodpeckernest cavities,Northern Flicker nest cavities, or both.
METHODS
[Auk, Vol. 107
were nesting, we censusednestson 8 plots in 1983
and 7 plots in 1984. Approximately 50% of the plots
censusedeachyear containedEuropeanStarlingnests.
Of the 7 plotscensusedin 1984,4 were censusedagain
in 1985 for a pairwise comparisonof the number of
nestspresent in 1984 and 1985. The 1985 censusdata
were not usedin any other analysisbecausethey were
not independent of the 1984 censusdata.
For each nest we found on or near the plots, we
measuredthe height and orientation of the cavity
entrance,the height of the saguaroin which the nest
was located (the nest saguaro),and the number of
arms on the nest saguaro.For nests that could be
reached with a 7.6-m ladder, we measured the vertical
and horizontal diametersof the cavity entrance,the
horizontaldepthof the cavity,and the verticaldepth
of the cavity (Fig. 1).
To samplethe vegetation,we selectedrandomly 10
pointsin every plot, and recordedall saguaroswithin
30 m of each point. For each saguaro,we estimated
height,andwe countedthe numberof armsandholes
that, from the ground, appeared to be possiblenest
cavities.A 60-m-long and 3-m-wide transectwas centered on each point and randomly oriented. We estimated the height and width of the foliage of all
trees, shrubs,and cacti (excepttriangle-leaf bursage
[Franseriadeltoidea]),which each transect intersected.
Only the numberof triangle-leafbursageintersected
by a transectwas recordedbecauseof the relatively
1,557-km 2 area in and around the Picacho Mountains,
uniform size of triangle-leaf bursage.We practiced
Pinal County, Arizona, and the Tucson Mountains, estimatingthe heights and widths of vegetation until
Pima County, Arizona. Saguarosin the study area we were accurateto within 30 cm, and we continually
usuallyoccurin large patchesthat reflectmicrocli- checkedour estimatesthroughout the study to mainmatic differences.We conductedpreliminary surveys tain this accuracy.
at eachaccessiblearea of saguarohabitat to determine
The volume of each plant intersectedby the tranwhether EuropeanStarlingswere present.We located sectswas calculated from the estimated height and
randomly 7 plots in areas of saguarohabitat with width of the plant. The shape of all plants except
splendens)
wasestimatedasa hemiEuropeanStarlingsand 8 plots in areasof saguaro ocotillo(Fouquieria
habitatwith no EuropeanStarlings.All plotswere at ellipsoid. For ocotillo,the plant shapewas estimated
least 600 m from any other plot, and the mean min- asa cone.The volume estimatesfor individual plants
were summedto provide an index of the volume of
imum distancebetween plots was 1,579 m.
We intensively searchedeachplot for severaldays each plant specieson each plot.
The plotswere delineatedon 1:2400U.S. Geological
until we were sure we had located all the European
Starling,Gila Woodpecker,and Northern Flickernests. Survey topographic maps. From these maps we meaThe visibility in saguarocacti forestsallowed us to sured the elevation, slope, aspect, and distance to
seebirds fly to and from their nests,and we could nearestagriculture for each plot. Agriculturewas deoften hear nestlingscalling from a nest.Only cavities fined as any area >-1 acre that was irrigated for the
with eggsor nestlingswere considerednests.When growth of vegetationnot typicalof the SonoranDeswe were uncertain if a cavity was a nest, we climbed ert. We defined agriculture to include areasof lawn
the saguarowith a ladderand lookedinto the cavity (such as golf courses,parks, and housing developwith a mirror and light. If we couldnot reacha suspect ments) becausethey are used by European Starlings
nestcavitywith a ladder,we checkedthe cavityevery for foragingmuch like farm fields(Dunnet 1955,Royweek until nestlingscould be heard, a bird was ob- all 1966, Troetschler 1976, Feare 1984, T. A. Kerpez
servedfeedingnestlings,or the nestingseasonended. pers. obs).
For all statisticaltests,the alpha level was0.05.The
The number of days each plot was censusedvaried,
depending on the number of saguarosand cavities alpha levels for multiple pairwise tests were calcuwe needed to examine. We finished censusingone lated as describedby Neu et al. (1974)to maintain an
alpha of 0.05 acrossexperiments. Means are given
plot before we started another.
Between8 April and 4 June,when all three species with standard errors.
We established15 square plots (10-ha each) in a
April 1990]
Starling
andWoodpecker
Competition
369
TABLE1. Heights and number of arms of saguaro
classes.
Class
Description
1
2
Height < 2.5 m
2.5 m -< height < 4.5 m
3
4.5 m _<height < 7.0 m & no. of arms < 6
4
Height >- 7.0 m or no. of arms >- 6
date the plot was censused.The number of European
Starlingnestson the plot alsowasusedas a possible
explanatoryvariable for the Gila Woodpeckerand
Northern Flickerregressions.
The numberof saguaro
holes that appearedto be possiblenest cavitieswas
also used as a possibleexplanatory variable for the
EuropeanStarling regression.
Fig. 1. Measurementsof cavity dimensions:vertical diameter of entrance (VDE), horizontal diameter
of entrance(HDE), horizontaldepth (HD), and vertical depth (VD).
Beforethe regressionanalysis,the dependentvariableswere transformedwith the squareroot transformation(•/x + 0.375)to causethe dependentvariables'meansto be independentof their variancesand
to makethe dependentvariablesmorenormally distributed (Draper and Smith 1981).Plotsof the regression residualsagainst the predicted values and the
independentvariableswere analyzedfor eachregression(asdescribedby Draperand Smith 1981)and met
the assumptionsfor multiple linear regressionanalysis.
To examine the relationship of plot aspectto the
numberof nestspresent,eachplot wasclassifiedinto
one of four 90ø quadratscentered on North, South,
Analysis
of factorsaffecting
thenumber
of nests.--We East,and West.We testeddifferencesamongquadrats
testedwith the Mann-Whitney testdifferencesin the with the Kruskal-Wallistest and a nonparametric
number of Gila Woodpeckerand Northern Flicker multiple comparison (Gibbons 1976). The effect of
nestson the plots with EuropeanStarlingsand the southernaspectcouldnot be testedbecauseonly one
plotswithout EuropeanStarlings.However, thesetests plot had a southern aspect.Differencesbetween 1984
did not accountfor habitat variablesthat may affect and 1985 in the number of nests present for each
the numberof nests;therefore,we usedmultiple lin- specieson the four plots censusedboth theseyears
ear regressionanalysiswith forward stepwise inclu- were testedwith the paired-samplet-test.
sion of variables(Dixon 1985, computerprogram
Anaiysis
ofnest-cavity
dimensions
andlocation.--We
BMDP-2R, F-to-enter = 4.0, F-to-remove = 3.9) to sitestedwhether the dimensionsof Gila Woodpecker,
multaneouslyexaminethe effectsof EuropeanStar- Northern Flicker,and EuropeanStarlingnestcavities
lings and habitat variables.Three regressionmodels differed by multivariate analysisof variance(overall
(onefor eachspecies)
were developed.The depen- differences),analysisof variance(differencesin each
dentvariables
werethenumberof nestsof eachspecies dimension),and the Student-Newman-Keulstest(difpresent on each plot. To include a measure of the
ferencesbetween species).Differencesin the nestsaguaroson the plotsas a possibleexplanatoryvari- cavity heights among the three specieswere tested
able, saguaroswere categorizedinto four classesthat with analysisof variance.Nest-cavityheight wasnot
were approximatelyequal in numberand easilydis- includedin the multivariateanalysisof variance,betinguished (Table 1), and the number of saguarosin causethe nestsfor which cavity dimensionswere
eachclasswas calculatedfor each plot. The possible measuredwere lower than the nestsfor which cavity
explanatoryvariableswere the following plot char- dimensions could not be measured. No variable's
acteristics:
the numberof saguaros
in eachof the four distribution differed significantlyfrom the normal
saguaro classes;the volume index for each of the 30
distribution(P > 0.05, Kolmogorov-Smirnovgoodtree,shrub,and cactispeciesfound on the plots;the ness-of-fittests;Zar 1984). Variables with signifitotalvolumeof all plant,tree,shrub,andcactispecies cantly unequal variances(P < 0.05, Bartlett-BoxF-test;
on eachplot;the distancefrom the plot to the nearest Nie et al. 1975)were transformedwith the logarithagriculture;the slopeof the plot; the elevationof the mic transformation[ln(x)] to equalizetheir variances.
plot; the year the plot was censused;and the mean
We testedwith the Rayleigh test (Batschelet1981)
370
whether
KERPEZ
ANDSMITH
the orientation
of nest cavities
was nonran-
dom for each species.Differencesbetween speciesin
their use of saguaroclasseswere tested with the binomia! test for two proportions(Zar 1984).
[Auk,Vol. 107
in 1985 had been usedby Gila Woodpeckersin
1984.
While censusingone of the plots, we observeda pair of Gila Woodpeckersfight with a
pair of EuropeanStarlingsfor a nestcavity(Kerpez 1986).When we returnedto thisplot a week
RESULTS
later, the nestcavity was occupiedby European
In 1983, 15 Gila Woodpeckernests,15 North- Starlingsand no Gila Woodpeckers
were nestern Flickernests,and 10 EuropeanStarlingnests ing on the plot.
were found on the 8 plots censused;and in 1984,
Factorsaffectingthe numberof NorthernFlicker
19 Gila Woodpeckernests,6 Northern Flicker nests.--The number of Northern Flicker nests
nests,and 11EuropeanStarlingnestswere found on the plots with EuropeanStarling nests(2?=
on 7 plots.
1.38 + 0.324, n = 7) was not significantly difFactorsaffecting
the numberof Gila Woodpeckerferent from the number of Northern Flicker
nests.--Thenumber of Gila Woodpeckernests nestson the plots without European Starling
on the plots with EuropeanStarling nests(2?= nests(2?= 1.43 + 0.571, n = 8) (P = 0.530). This
0.86 + 0.404, n = 7) was significantly lessthan was confirmed by the regression analysis.
the number of Gila Woodpeckernests on the Throughoutthe stepwiseinclusionof variables
plots without European Starling nests(2?= 3.5 in the Northern Flicker regression,the number
+ 0.707, n = 8) (P = 0.01). This was confirmed of EuropeanStarlingnestswasnever significant
by the regressionanalysis.Of all the explana- in explaining the number of Northern Flicker
tory variables examined, the number of Euro- nests (P -> 0.927).
pean Starling nests explained the most variaThe year the plot was censusedexplained
tion in the number of Gila Woodpeckernests. 25.0% of the variation in the number of NorthThe relationshipof EuropeanStarling neststo ern Flicker nests (P = 0.0151, b = -0.5436). BeGila Woodpeckernestswas negativeand alone causeyear was codedas i = 1983and 2 = 1984,
explained46.7%of the variation in the number the negativecoefficientin the regressionequaof Gila Woodpeckernests (P = 0.0012, b = tion means that there were fewer Northern
-0.1704). The number of class4 saguaros,the Flicker nests in 1984 than in 1983. There also
largestsaguaros(Table 1), waspositivelyrelated were significantlyfewer Northern Flickernests
to the numberof Gila Woodpeckernests.After in 1984 than in 1985on the four plots censused
the relationshipto EuropeanStarlingnestswas both years (P = 0.007). Examinationof winter
determined, the number of class4 saguarosex- precipitation recordedat two weather stations
plained an additional 18.1%of the variation in near the study area (Tucsonand Eloy) showed
the number of Gila Woodpeckernests (P = that there was much less precipitation in the
0.0167,b = 0.1593).The slope of the plot was period December-Marchbeforethe 1984breednegativelyrelatedto the numberof Gila Wood- ing season(5.1 cm) comparedwith the precippeckernestsand explainedan additional16.2% itation before the 1983 (22.0 cm) and 1985 (17.2
of the variation
in the number of Gila Woodcm)breedingseasons
(NationalOceanicandAtpecker nests(P = 0.0109,b = -0.0434). Togeth- mospheric Administration 1982, 1983, 1984,
er, thesethree variablesexplained81.0%of the 1985).
variation in the number of Gila Woodpecker
The volume of ironwood (01neyatesota)was
nests(P = 0.0003,intercept = 1.643).
positively related to the number of Northern
There were no significantdifferencesin the Flicker nests.After the effect of year was denumber of Gila Woodpeckernestson plotswith termined, ironwood volume explained an addifferent aspects(P = 0.917). There was no sig- ditional 20.7% of the variation in the number
nificant
difference
between
1984 and
1985 in
the number of Gila Woodpeckernestson the
four plotscensusedboth years(P = 0.240).How-
of Northern Flicker nests(P = 0.0539,b = 0.0483).
The volumes of the following plants found on
the plotswere significantlycorrelatedwith the
ever, one of the plots contained 3 Gila Wood- volume of ironwood: gray-thorn (Condalialypecker nestsand 0 EuropeanStarling nestsin cioides)(r = 0.917, P = 0.001), deserthackberry
1984. This plot contained2 EuropeanStarling (Celtispallida)(r = 0.932, P = 0.001), burro-bush
nestsand only i Gila Woodpeckernestin 1985. (Hymenoclea
salsola)(r = 0.940, P = 0.001), cane
The 2 nest cavitiesused by EuropeanStarlings cholla (Opuntiaspinosior)
(r = 0.944, P = 0.001),
April1990]
Starling
andWoodpecker
Competition
371
TABLE
2. Dimensions(cm) and height (m) of Gila Woodpecker,EuropeanStarling,and Northern Flicker
nest cavities.Differencesamongspecieswere testedwith ANOVA and the Student-Newman-Keulstest.
Meansfollowed by the sameletter did not differ significantly(P > 0.05).
Nest cavityvariable
Gila Woodpecker EuropeanStarling
(• + SE)
(œ+ SE)
Entrance vertical diameter 2
Entrance horizontal diameter 2
Cavityverticaldepth2
Cavity horizontaldepth2
Height of nest3
From
Northern Flicker
(œ+ SE)
P•
0.0023
<0.0001
5.7 + 0.18 A
6.3 + 0.21 A
5.7 + 0.22 A
6.6 + 0.25 A
7.0 + 0.46 B
8.3 + 0.42 B
27.8 + 0.99A
15.7 + 0.63A
5.8 + 0.14A
31.8 + 2.02A
14.0 + 0.93AB
6.0 + 0.19A
37.6 + 1.45B
12.5+ 0.97B
6.2 + 0.23A
0.0001
0.0252
0.3036
ANOVA.
n = 32 for Gila Woodpeckers,n = 19 for EuropeanStarlings,and n = 15 for Northern Flickers.
n = 64 for Gila Woodpeckers,n = 26 for EuropeanStarlings,and n = 28 for Northern Flickers.
and honeymesquite(Prosopis
julifiora),(r = 0.533,
P = 0.041).Together,the year the plot wascensusedand the volume of ironwood explained
45.7% of the variation
in the number
of North-
ern Flicker nests(P = 0.0257,intercept = 1.958).
There were no significantdifferencesin the
number of Northern Flicker nestson plots with
different aspects(P = 0.726).
Nest-cavity
dimensions
andlocation.--Throughout thestudy,64 Gila Woodpecker,
28 Northern
Flicker, and 26 EuropeanStarling nestswere
found on or near the plots. We measuredthe
cavity dimensionsof 32 Gila Woodpecker,15
Northern Flicker, and 19 European Starling
nests.EuropeanStarlingsand Gila Woodpeckers did not significantly differ in nest-cavity
Factors
affecting
thenumber
of European
Starling dimensions(P = 0.3840).EuropeanStarlingsand
Gila Woodpeckersboth significantlydiffered
nests.--Themean number of EuropeanStarling
nestson the plotswith EuropeanStarlingswas
3.00 + 0.926 (n = 7). The distancefrom the plot
to the nearestagriculturewasnegativelyrelated
to the number of EuropeanStarling nestsand
alone explained 29.2% of the variation in the
numberof EuropeanStarlingnests(P = 0.0078,
b = -0.2260). The mean date the plot was censusedwas negativelyrelated to the number of
European Starling nestsand explained an ad-
from Northern Flickers in nest-cavity dimensions(P = 0.0182and P = 0.0001,respectively).
EuropeanStarling and Gila Woodpeckernest
cavities had significantly smaller entrances
(verticaland horizontal diameters),and the cavitiesweresignificantlyshallowerin the vertical
planeof the saguaro
thanNorthernFlickernest
cavities(Table 2). The three speciesdid not sig-
nificantlydifferin nest-cavityheight(Table2).
EuropeanStarlingsand Gila Woodpeckers
did
of EuropeanStarling nests(P = 0.0208, b = not differ significantlyin their useof saguaros
-0.2319). Together, these two variables ex- for nestsites(P = 0.511)(Fig. 2). EuropeanStarplained 55.5%of the variation in the number lings and Gila Woodpeckersnestedin class3
significantlylessand in class4 saguaof EuropeanStarlingnests(P = 0.0078,intercept saguaros
= 2.220).
ros significantlymore than Northern Flickers
Plotswith northern aspectshad significantly (P _<0.013)(Fig. 2). The orientationof nestcavfewer EuropeanStarling neststhan did plots ities did not significantly differ from random
with easternaspects(P = 0.027).All three plots for Gila Woodpeckers(r = 0.02, P > 0.90),
ditional
26.3% of the variation
in the number
with northern aspectshad no EuropeanStarling
nests;however, this was probably not caused
by their northern aspectbut insteadby their
distancefrom agriculture.The three plotswith
northernaspectswere all >4 km from agriculture. The farthestplot from agriculturewith a
EuropeanStarlingnest was ca. 4 km from ag-
Northern Flickers (r = 0.16, P > 0.47), and Eu-
ropeanStarlings(r = 0.21, P > 0.30). For more
information on Gila Woodpeckerand Northern
Flicker nest-siteselectionand nest-cavitycharacteristicsseeKerpez and Smith (1990).
DISCUSSION
riculture.
There was no significantdifferencebetween
1984 and 1985 in the number of European Star-
ling nestson the four plotscensusedboth years
(P = 0.392).
Competition
between
European
StarlingsandGila
Woodpeckers.--European
Starlingscompetewith
Gila Woodpeckersfor nest cavitiesin saguaros
and this competitiondecreasesthe number of
372
KwP•ZANDSMITH
[Auk,Vol.107
excavatecavities only during winter and the
lossof a nestcavityduring springmay not trigger them to excavateanother one. Soule (1964)
saw Gila Woodpeckersexcavatecavities in February. There are no other published reportson
the time of Gila Woodpeckercavityexcavation.
The Museumof Ornithology at the University
of Arizona and the TucsonAudubon Society
• 40
CLASS 3 SAGUAROS
CLASS 4 SAGUARO•
Fig.2. Useof saguaros
for nestsitesby Gila Woodpeckers (n = 64), Northern Flickers (n = 28), and
European Starlings (n = 26). Class 1 and class2 sa-
guaroswere not used for nest sites by any of the
species.
Gila Woodpeckersthat nest in areaswhere European Starlingsnest. EuropeanStarlingsnest
in Gila Woodpeckernest cavities,and no other
factorsexaminedcould explain the negativerelationship between the number of European
Starling nestsand the number of Gila Woodpecker nests.
EuropeanStarlingsuse cavitiesin saguaros
only when they nestand are rarely seenin saguaro habitat during the rest of the year (T. A.
Kerpez pers. obs.).Plots that had many European Starling nests and no Gila Woodpecker
nests during the breeding seasonhad no EuropeanStarlingsand severalGila Woodpeckers
presentduring the rest of the year. Apparently,
at the beginning of the nestingseasonEuropean
Starlingsmove into areasthat Gila Woodpeckers inhabit the rest of the year and usurp nest
cavitiesfrom the Gila Woodpeckers.
Gila Woodpeckersthat losetheir nestcavities
to European Starlings did not excavateanother
cavity and nest in the same area. If they did,
there would be more Gila Woodpeckersthat
nest in areas where European Starlings nest.
There are several possible reasonswhy Gila
Woodpeckersdo not excavateanother cavity in
the samearea. There may be a lack of suitable
sitesfor cavities.Many of the holesin saguaros
are only a few inchesdeep (T. A. Kerpez pers.
obs.).Woodpeckersmay begin excavatingcavities only to find that the site is unsuitable.It
is also possiblethat EuropeanStarlings harass
Gila Woodpeckersif the woodpeckersremain
in the area.
receive almost all of their calls complaining
aboutGila Woodpeckersexcavatingcavitiesin
buildings during winter (S. M. Russellpers.
comm.). For three years we observed Gila
Woodpeckersalmostdaily during the breeding
season(April-June), and we never saw them
excavatecavities.Because
Gila Woodpeckers
nest
in the same cavity for several years (Bendire
1892,Gilman 1915,S. M. Russellpers. comm.,
Kerpez 1986), they may excavatea new cavity
only when necessary.EuropeanStarlingswere
absent during winter; therefore, Gila Woodpeckersmay not be stimulatedto excavatemore
cavities during winter.
CompetitionbetweenEuropeanStarlingsand
Gila Woodpeckers
couldaffectthe entirecavitynestingcommunity.The two plotswith the most
European Starling nests had no Gila Wood-
pecker nests. Communitieswith many European Starlingsmay no longer have Gila Woodpeckers, and communities with intermediate
numbersof EuropeanStarlingsmay havefewer
Gila Woodpeckers.If Gila Woodpeckerscannot
nestin an area,they may not excavatenew cavities in that area. If this is true, as the saguaros
with existing cavities die, the number of cavities available
will
decrease in areas where
Gila
Woodpeckersare excludedor reducedin number by EuropeanStarlings.This decreasein the
numberof availablecavitieswould be largebecauseGila Woodpeckersand Northern Flickers
are the only common excavatorsof cavitiesin
saguaros,and exceptwhere EuropeanStarlings
were present, Gila Woodpeckers were much
more numerous
in the areas we studied than
Northern Flickers. A decrease in the number of
cavitiesavailable could have a profound effect.
There are six other speciesof native birds that
regularly nest in cavitiesin saguaros:Elf Owls
(Micrathenewhitneyi),Brown-crestedFlycatchers (Myiarchustyrannulus),Ash-throatedFlycatchers(M. cinerascens),
Purple Martins, Western Screech-Owls
(Otus kennicottii), and
American Kestrels(Falcosparverius)
(Bent 1937,
Another possibilityis that Gila Woodpeckers 1942; Allen and Nice 1952). These birds do not
April1990]
Starling
andWoodpecker
Competition
373
excavatecavitiesbut depend on the woodpeck-
study area in 1984 than in 1983 and 1985.This
er's cavities. Also, if there are fewer Gila Wood-
wasprobablycausedby thelackof precipitation
pecker cavities available, European Starlings
during the months preceding nesting (Decemmay competewith Northern Flickersfor nest ber-March) in 1984 compared with 1983 and
cavities.Also, EuropeanStarlingsmay compete 1985. Northern Flickers in the Sonoran Desert
with the other secondarycavitynesters,if they forageprimarily for insectson the groundand
do not already.
in annual foliage <10 cm high (Tomoff 1974,
EuropeanStarlingsand NorthernFlickers.--Eu- Vander Wall 1980). In desertsthe timing and
ropeanStarlingscompetewith Gila Woodpeck- quantityof precipitationplaysan importantrole
ers and not with Northern Flickers probably in the germination and growth of annuals
becauseEuropeanStarlingsmay be able to dis- (MacMahon and Schimpf 1981). The germinaplace Gila Woodpeckers more easily than tion and growth of annuals along with soil
Northern Flickers. Northern Flickers are larger moistureprobablyaffectthe productionof inthan Gila Woodpeckers(Ridgeway 1914, Dun- sects that Northern Flickers eat and feed their
ning 1984),and they may alsobe more aggres- nestlings.
sive.Northern Flickersusurpnestcavitiesfrom
Gila Woodpeckerswere not affected by the
Gila Woodpeckers(Brenowitz 1978, Martindale
lackof precipitationin 1983becauseduring the
1982), but the reverse situation has not been nestingseasonthey forageprimarily on saguaro
reported.
cacti(Martindale 1983).Seasonaldrought does
Habitatfactorsaffecting
thenumber
ofGilaWood- not affectthe productionof flowersand fruits
peckernests.--The number of class4 saguaros by saguarosbecausesaguarosstore water repositively affectsthe number of Gila Wood- serves in their succulent stem tissue (Steenpeckersthat nest in an area becauseGila Wood- berghand Lowe 1977).EuropeanStarlingswere
peckersuse then for nestingand foraging. Gila not affectedby the lackof precipitationin 1983
Woodpeckersnest almostexclusivelyin class4 because,during the nesting seasonin the Sosaguaros(Fig. 2; Kerpezand Smith 1990).Dur- noran Desert,they forageprimarily for insects
ing the nestingseason,Gila Woodpeckersspend in irrigated agricultural areas (Royall 1966, T.
more than half their foraging time on saguaros, A. Kerpez pers. obs.), which produce insects
feedingtheir nestlingspollen, fruit, and insects relatively independent of seasonalprecipitagleanedfrom saguaros(Martindale 1983).Class tion.
The volume of ironwood was related posi4 saguaroshave the mostflowersand fruits, are
the tallest saguaros,and have the most arms tively to the numberof Northern Flickersthat
(Steenberghand Lowe 1977). Also, class4 sa- nest in an area probably becausethe presence
guaroshave the mostsurfacearea from which of ironwood indicates a warmer microclimate
Gila Woodpeckersglean insectsand may have (Kearnyand Peebles1960)and a differentvegemore insectsper unit area becauseclass4 sa- tation community.The volumes of five plant
guarosare older.
specieswere correlatedsignificantlywith the
Slopeis negativelyrelated to the number of volume of ironwood. The warmer microclimate
Gila Woodpeckersthat nest in an area possibly and the vegetation community indicated by
becauseit affectsthe vegetation. Differences in ironwood may produce more insects that
the totalvolumeof all plant, tree,shrub,or cacti Northern Flickers eat and feed their nestlings.
species,
or the volumeof any singleplantspecies
The numberof class4 saguarosdid not affect
could not explain the relationship that slope the numberof Northern Flickersnestingon the
had to the number of nesting Gila Woodpeck- plots becauseall the plots had some class4 saers. Steepslopesmay causedifferencesin the guaros.Northern Flickersrarely use saguaros
amountof severalplant speciesthat are not ob- for foraging (Tomoff 1974,Vander Wall 1980),
viouslyrelated.Thesedifferencesmay affectthe and one large saguarois probably sufficientfor
availabilityof insectsthat Gila Woodpeckers
eat each nesting pair. Further, Northern Flickers
and feed their nestlings.Also,steepslopesmay often useclass3 saguarosfor nest sites(Fig. 2;
increasethe energyrequired to forageand de- Kerpez and Smith 1990).
liver food to a nest.
Factorsaffectingthenumberof European
Starling
Factorsaffectingthe numberof NorthernFlicker nests.--The distanceto agriculture was correnests.--Fewer
Northern
Flickers nested on the
lated negatively with the number of nesting
374
KERI'EZ
ANDSMITH
EuropeanStarlingsin an area. EuropeanStarlings that nest in the Sonoran Desert obtain
mostof their food from agriculturalareas(Royall 1966, T. A. Kerpez pets. obs.).During our
census,we saw no European Starlings forage
on the plots.EuropeanStarlingsalwaysflew off,
usuallytowardagriculture,to forage.When we
were closeenough to seeagricultural areas,we
saw EuropeanStarlings repeatedly leave their
nest,fly to the agriculture,land on the ground,
and return
to the nest with
insects.
Other
re-
ports confirm that during the nestingseason
EuropeanStarlingsforageprimarily in agricul-
[Auk,Vol. 107
ACKNOWLEDGMENTS
We thank R. William Mannan for hisadviceduring
the studyand for criticallyreviewingthe manuscript,
and William J. Matter for critically reviewing the
manuscript.Bruce D. Leopold and Robert O. Kuehl
provided invaluable assistancewith the statistical
analysis.We are especiallythankful to BarbaraSunshine for her supportand help throughoutthis study
and for typing the manuscript.This study was funded by the Arizona CooperativeWildlife ResearchUnit,
the University of Arizona, and the Arizona Game and
Fish Department.
tural areas (Dunnet 1955, Royall 1966, Troet-
LITERATURE CITED
schler 1976, Feare 1984). Feare (1984) found that
agricultural areas supported the highest den- ALLEN,R. W., & M. M. NICE. 1952. A study of the
breeding biology of the Purple Martin (Progne
sitiesof breedingEuropeanStarlingsin Europe.
subis).Am. Midi. Nat. 47: 606-665.
European Starlings nested on the plots
BATSCHELET,
E. 1981. Circular statisticsin biology.
throughout the censusperiod, but fewer EuNew York, Academic Press Inc.
ropeanStarlingsnestedtoward the end of the BENDIRE,C.E. 1892. Life histories of North American
censusperiod. In Arizona, many EuropeanStarbirds with specialreference to their breeding
lings have two broodsand begin their second
habitsand eggs.U.S. Natl. Mus. Bull. I.
brood in May (Royall 1966).Usually fewer Eu- BENT, A. C. 1937. Life histories of North American
ropeanStarlingshavesecondbroodsthan have
birds of prey, part 2. U.S. Natl. Mus. Bull. 170.
1939.
first broods (Kessel 1957). The decreasein the
number of EuropeanStarlingsnesting on the
plots may be due to single broods.
Future implicahons.--Competition between
EuropeanStarlingsand Gila Woodpeckerswill
probablybecomemore severeand more widespread.From 1968 to 1976, the number of European Starlingsin the southwesternUnited
States has more than doubled (Dolbeer and
Life
histories
of
North
American
woodpeckers.U.S. Natl. Mus. Bull. 174.
1942. Life historiesof North American flycatchers, larks, swallows and their allies. U.S. Natl.
Mus.
--.
Bull.
179.
1950. Life historiesof North Americanwagtails, shrikes, vireos and their allies. U.S. Natl.
Mus.
Bull. 197.
BRENOWITZ,
G. L. 1978. Gila Woodpeckeragonistic
behavior.
Auk 95: 49-58.
Stehn 1979). EuropeanStarlingsare probably DIXON, W. J. (Ed.). 1985. BMDP statistical software,
1985 ed. Berkeley, Univ. California Press.
still increasingin. the SonoranDesert, but it is
R. A., &: R. A. STEHN. 1979. Population
difficult to tell if they will increaseonly near DOLBEER,
trendsof blackbirdsand starlingsin North Ameragriculture.We did not find EuropeanStarling
ica 1966-1976. U.S. Fish Wildl. Serv., Spec. Sci.
nestson plots farther than 4 km from agriculture. However, if European Starlings are in-
Rep. Wildl. 214.
DRAPER,
N. R., & H. SMITH. 1981. Applied regression
creasing,they maybeusingonly their preferred
analysis,2nd ed. New York, JohnWiley & Sons.
habitat (i.e. areasnear agriculture)now. Once DUNNET,G. M. 1955. The breeding of the Starling
the preferredareasare full, EuropeanStarlings
Sturnusvulgarisin relation to its food supply.Ibis
97: 619-662.
mayinvadeareasfarther from agriculture.Also,
human activity has decreasedthe amount of DUNNING,J.G., JR. 1984. Bodyweightsof 686 species
of North American birds. West. Bird Band. Assoc.
desertthat is far from large lawns and agriculMonogr. I.
ture. If EuropeanStarlingscontinueto increase
and spread,the survival of Gila Woodpeckers FEAR•,C. 1984. The starling.New York,Oxford Univ.
could be threatened.
A decline
in Gila Wood-
Press.
GIBBONS,J. D. 1976. Nonparametric methods for
quantitative analysis.New York, Holt, Rienhart,
peckers could profoundly affect the unique
communityof birdsthat nestin saguarocavities
& Winston.
and could also affect the survival of saguaros GILMAN,M. F. 1915. Woodpeckersof the Arizona
lowlands. Condor 17: 151-163.
becauseGila Woodpeckersmay be important
pollinators of saguaros (S. Martindale pets. HOWELL,
A.B. 1943. Starlingsandwoodpeckers.
Auk
comm.).
60: 90-91.
April 1990]
Starling
andWoodpecker
Competition
KEARNEY,
T. H., & R. H. PEEBLES.
1960. Arizona flora,
2nd ed. Berkeley,Univ. of California Press.
KERPEZ,
T.g. 1986. CompetitionbetweenEuropean
Starlingsand native woodpeckersfor nestcavities in saguaros.M.S. thesis, Tucson, Univ. Arizona.
--,
& N. S. SMITH. 1990. Nest-site selection and
nest cavity characteristics
of Gila Woodpeckers
and Northern
Flickers. Condor 92: 193-198.
KESSEL,
B. 1953. Distributionand migrationof the
EuropeanStarling in North America.Condor 55:
--.
375
NIE, N. H., C. H. HULL, J. G. JENKINS,K. STEINBRENNER,
& D. H. BRENT.1975. SPSS,statisticalpackage
for the social sciences, 2nd ed. New York,
McGraw-Hill.
NILSSON, S. G.
1984. The evolution
of nest-site se-
lection among hole-nesting birds: the importanceof nest predationand competition.Ornis
Scandinavica
15: 167-175.
PETERSON,
g., & G. GAUTHIER.1985. Nest site useby
cavity-nesting birds of the Cariboo Parkland,
British Columbia.
Wilson
Bull. 97: 319-331.
49-67.
RIDGEWAY,R. 1914. The birds of North and Middle
1957. Astudy of the breedingbiologyof the
European Starling (Sturnusvulgaris)in North
ROYALL,
W. C., JR. 1966. Breedingof the starlingin
America.
Am. Midi.
Nat.
58: 257-331.
America.
U.S. Natl.
central Arizona.
Mus. Bull. 50.
Condor
68: 196-205.
KILHAM,L. 1958. Pairformation,mutualtappingand SHELLY,
L. O. 1935. Flickersattackedby starlings.
Auk 52: 93.
nestholeselectionof Red-belliedWoodpeckers.
Auk 75: 318-325.
SOULE,O. 1964. The saguarotree-hole microenviMACMAHON, J. A., & D. J. SCHIMPF. 1981. Water as
a factorin the biology of North American desert
ronment
in southern
Arizona.
II: Summer.
M.S.
thesis, Tucson, Univ. Arizona.
plants.Pp. 114-171in Waterin desertecosystems STEENBERGH,
W. F., & C. H. LOWE. 1977. Ecologyof
(D. EvansandJ.Thames,Eds.).U.S./I.B.P.Synth.
the saguaroII. Natl. ParkServ.Sci.Monogr.SeSeries
12.
ries 8.
MARTINDALE,
S. 1982. Nestdefenseandcentralplace
foraging:a modeland experiment.Behav.Ecol.
Sociobiol.
10: 85-89.
1983. Foraging patterns of nesting Gila
Woodpeckers.Ecology64: 888-898.
MCGILvRE•,F. B., & F. M. UHLER. 1971. A starling
deterrentWoodDuck nestbox.J.Wildl. Manage.
35: 793-797.
MONSON,G. 1948. The starling in Arizona. Condor
50: 45.
NATIONAL
TOMOFF,
C.S. 1974. Avian speciesdiversityin desert
scrub.Ecology55: 396-403.
TROETSCHLER,
T.G. 1976. AcornWoodpeckerbreeding strategyasaffectedby starlingnest-holecompetition. Condor 78: 151-165.
VAN BALEN,J. H., C. J. H. BOOY,J. A. VAN FRANEKER,
& E.R. OSIECK.
1982.Studieson hole-nestingbirds
in naturalsites:i. availabilityand occupationof
natural
nest sites. Ardea 70: 1-24.
VANDER WALL, S. B. 1980. The structure of the SoOCEANIC AND ATMOSPHERIC ADMINISTRA-
TION. 1982,1983,1984,1985. Climatological
data:
Arizona, annual summary.Asheville, North Carolina, Natl. Climatic Ctr.
NEU, C. W., C. R. BYERS,
& J. M. PEEK. 1974. A tech-
nique for analysisof utilization-availabilitydata.
J. Wildl. Manage. 38: 541-545.
noran Desertbird communities:effectsof vegetation structure and precipitation. Ph.D. dissertation, Logan, Utah State Univ.
ZAR,J. H. 1984. Biostatistical
analysis,2nd ed. EnglewoodCliffs, New Jersey,Prentice-Hall,Inc.