SEASONAL
CHANGES
DISTRIBUTION
IN THE
HABITAT
OF TRANSIENT
INSECTIVOROUS
BIRDS
SOUTHEASTERN
IN
ARIZONA:
COMPETITION
MEDIATED?
RICHARD L. HUTTO
Departmentof Zoology,Universityof Montana,Missoula,Montana49812 USA
ABSTRACT.--The
distribution and abundanceof 26 migratory insectivorousbird species
were recorded over an elevational habitat gradient in the Chiricahua Mountains, Arizona
for the spring and fall migratory seasons.Most of the speciesused this area only during
migratory passage,and 54% exhibited significantshifts in the habitatsoccupiedfrom spring
to fall. The majority (69%) of speciesalso exhibited significantchangesin density within
habitatsbetween seasons.Using pairwise correlationsof bird densitiesfrom 7 habitat types
and 2 seasons,I identified 5 groups that contained specieswhose seasonaldistributional
patterns were similar to one another but independent and distinct from members of the
other 4 groups. Despite independence among groups in the seasonalpatterns of habitat
distribution,the combineddensityof all specieswassignificantlypositivelycorrelatedwith
a measureof food availability taken from eachof the habitat types in eachmigratory season.
Consequently,the spring-to-fall changein insectdensity within each habitat also was significantlycorrelatedwith the seasonalchangein bird densityover eachof the habitattypes.
The hypothesesthat best explain these correlationsinclude that in which competitive adjustmentsamong the migratory birds enable a closematch to food resourceavailability and
that wherebynoncompetitiveadjustmentsoccurin responseto the diversity(itselfcorrelated
with food abundance)of food typesavailable.Received
7 May 1984,accepted
18 August1984.
EXPLANATIONS
for patternsin bird community
structure routinely have been sought from
knowledge of food resourcedistributionsand
abundances(MacArthur 1969, 1970, 1972;Cody
1974). This stems logically from the idea that
natural
selection
should
lead toward
the utili-
zation of currently underutilized resources,and
competitionamong speciesshould determine
both
the number
and
relative
abundance
of
ministicones(Wiens 1977, 1981;Rotenberryand
Wiens 1980a, b).
Although most of the discussionand uncertainty about the role of food resourcelevels in
controlling bird community composition revolves around breeding populations,a unique
situation for testing whether birds respond,in
an ultimate sense,to seasonalchangesin food
levels existswith migratory birds. Changes in
vegetation structure and food resource levels
speciesthat use those food resources.More recently, however,Wiens (1977, 1983) suggested from spring to fall are not of equal magnitude
that the match between
resource levels and
among habitat types. This provides the opporspecies utilization patterns will be less than tunity to ask whether bird community strucperfect becauseunpredictability or instability ture changesin responseto changesin the disin resource levels cannot be tracked rapidly tribution of food resources.
enough by bird populations (but see Cody
In this paper I outline the patterns of habitat
1981). In fact, recent experimental evidence use by small, insectivorous bird species of
(Emlen 1978, 1979, 1981; Emlen and DeJong southeasternArizona during both migratory
1981)suggests
that birdsare ratherfixedin their seasons and test whether the habitat distribuforagingbehaviorsand that a closetrackingof tion of thesebirds is independentof migratory
resourcelevels is impossible. This means that, season.This is followed by an analysisof the
except for the occasionalyear of an ecological relationships between bird densities and varcrunch when resources are scarce, many as- ious habitat parameters,including food availpects of community structure would result as ability. If interspecificand intraspecificcommuch from stochasticprocessesas from deter- petition for food are important during the
120
The Auk 102:120-132.January1985
January
1985]
Habitat
Selection
byMigratory
Birds
121
migratoryseasons,
then the combineddensity
Bird censuses.--Icensused birds in each site by
walking a l-kin line transectbeginning at daybreak
expectedto match the relative food resource and recordingbirds detectedby sight or soundwith-
of similar-sized, insectivorous birds would be
levels among habitats within a seasonand,
therefore,to matchany seasonal
changesin relative food levels amonghabitats.If food is of
little ultimate importancein determiningbird
in a fixed width
that varied from 25 to 30 m, de-
pending upon the habitat involved. The fixed-width
transectmethod provides bird density estimatesthat
are thought to be quite reliable relative to other commonly employed transectcensustechniques(Am-
communitycomposition,or if food levelsare man and Baldwin 1960, Robinette et al. 1974, Franzreb
importantbut changesin food levels are im- 1981,Tilghman and Rusch1981).At least4 censuses
possibleto track closely, then independence were conductedin each site during each season(a
between
bird
densities
and food
densities
would be expected.
number deemedadequatefor comparativework; Anderson and Ohmart 1977). For each site the 1975 censuses were combined
STUDY AREA AND METHODS
I conductedbird censusesand recordedvegetation
parametersand insectdensitiesin eachof 7 homogeneoushabitat types in the Chiricahua Mountains,
Arizona (Fig. I). The 7 sitesoccurredalong an elevationalgradientand rangedfrom low and simple
with
1978 censuses for the fall
sample. In all but 2 cases,a single 1975 censuswas
combined
with
3 or more
1978 censuses.
In the
2
exceptions(desert flats and pine-oak woodlands), 2
censuses were conducted in 1975 and the rest in 1978.
Becausethe distribution of bird densities among
speciesdid not differ betweenyears(ANOVAs, P >
0.05), I feel that combining fall censusdata from the
two years is justified.
Although the same censusroute within a site was
used each time, for statisticalpurposesI treated each
censusas an independent estimateof the bird density in a given habitat because(I) stopoverperiods
for transients(the largemajorityof birdsin this study;
Prosopis
julifiora.The secondsite (desertwash) was seeResults)rarely exceeded4-6 days(unpubl. bandlocatedalong lower Cave Creek 2 km northeastof ing records)and (2) successivecensusesin a given
Portal (31ø55'N,I09ø07'W)at an elevation of 1,433m
habitat were spacedat least one week apart due to
and containedmany of the vegetationelementsthat the rotation of censusesamong sites.
characterizedthe desert flats plus scatteredPlatanus
For the purposesof this report I have restrictedmy
wrightiiand dense sectionsof Fallugiaparadoxaand analysesand discussionto the small, insectivorous,
Chrysothamnus
nauseosus.
The third site(creekbottom) foliage-gleaningbird speciesthat belong to the famwas locatedalong the south fork of Cave Creek 7 km
ilies Remizidae, Aegithalidae, Muscicapidae(Sylvisouthwest of Portal (31'52'N, 109*11'W) at an elevainae only), Vireonidae, and Emberizidae (Parulinae
tion of 1,631 m; this site was dominated by Pinus only; A.O.U. 1983).
Bird residence
status.--A bird specieswas classified
ponderosa,
Quercus
arizonica,Q. hypoleucoides,
and Picea
engelmanni.
The fourth site (pine-oakwoodland)was as transientin a given study site if it could be found
there only during one or both migratory seasons.The
located behind the American
Museum's Southwestbreeding status of each speciesin each site was deern Research Station 7 km southwest
of Portal
(31'53'N, I09ø12'W) at an elevation of 1,676 m and termined from breedingcensusdataprovided by Balda (1967),M. Cody (pets.comm.),and K. Garrett(pers.
to tall and complexin vegetationstructure.The first
site (desertflat) was located2 km northeastof Portal
(31055'N, 109ø07'W)at an elevation of 1,402 m and
was dominated by desertscrubvegetation, including
Acaciaconstricta,
Larreatridentata,Chilopsis
linearis,and
was dominatedby Pinusleucophylla,
Quercusemoryi,
comm.), each of whom conducted censusesclose to,
Q. arizonica,
andJuniperus
deppeana.
The fifth site(pineif not precisely within, my study sites.
oak-juniper woodland) was located 12 km west of
Vegetationmeasurements.--I
measuredsome vegePortal (31'57'N, 109'16'W) at an elevation of 2,286 m
tation parametersthat have been shownthrough preand was dominatedby the samevegetationelements vious work (James 1971, Whitmore 1975) to be imas the pine-oak woodland, plus Pinuscembroides,
P. portant in distinguishingbird speciesthat co-occur
ponderosa,
and Quercus
gambelii.The sixth site (pine within a restrictedgeographiclocation. On a single
forest) was located at the turnoff to Barfoot Park 12
km west of Portal (31055'N, I09ø16'W) at an elevation
occasionin each site during the fall of 1975 and the
spring of 1976,I countedthe number of times foliage
of 2,512 m and was dominated by Pinusponderosa. hit an extendablepole that was raised through the
The last site (pine-fir forest) was located at Rustler vegetationat 100 points (I every 10 m) 5 m to one
Park 13 km west of Portal (31ø54'N, I09ø17'W) at an
side of the censusroute, alternating left and right
elevation of 2,682 m and was dominated by Abies from one point to the next. An imaginary extension
concolor,
Piceaengelmanni,
Pseudotsuga
menzeisii,
Pinus for taller habitatswas provided by a cameraand teleponderosa,
and P. strobiformis.
photo lens. I usedthe total number of hits as an es-
122
RICHARD
L. HUTTO
[Auk,Vol.102
timate of vegetation density (VEGTOT), the total
number of hits from ground level to 1 m as an estimate of understory density (VEGONE), the proportion of points that containedat least 1 hit at >5 m as
an estimateof canopy cover (CANCOV), the number
of perennial plant specieshit as an estimateof plant
speciesrichness(PSRICH), and the maximum vegetation height at each point averagedover all points
as an estimateof mean vegetationheight (HEIGHT).
Three vegetationvariableswere statisticallysignificantly (P < 0.05) intercorrelated:VEGTOT with
CANCOV (r = 0.78), VEGTOT with HEIGHT (r =
0.74), and CANCOV with HEIGHT (r = 0.96). I re-
tained all variablesin my analyses,however, because
they were different enough to reveal differences in
the significanceof their correlationswith bird density. None of the vegetation variables was significantly correlatedwith the index of insectabundance
(defined below).
Fig. 1. Locationsof the 7 study sitesin the Chiricahua Mountains, southeasternArizona. (1) Desert
flat, (2) desertwash, (3) creek bottom, (4) pine-oak
woodland,(5) pine-oak-juniperwoodland,(6) pine
forest,(7) pine-fir forest.
Foodavailability.--I did not attempt to sample the
same prey speciesthat foliage-gleaninginsectivores
capture.Even samplesthat include only prey species
captured by the birds would not be immune from
criticismthat the prey cannotbe capturedin the same
RESULTS
manner that birds capture them and, therefore, that
such samplesstill might fail to provide an accurate
measureof food availability. Instead, I used a more
A total of 26 small, insectivorous,
foliagegeneral sampling scheme (below) and made the asgleaning bird specieswas recorded from the
sumption that the calculatedvalues were correlated spring and fall migration-periodcensuses
(Tawith actual prey availability.
ble 1). With the exceptionof the creek-bottom
Prey availability was estimated from counts of
habitat,the number of bird speciesin eachof
flying insectscaught on 10 x 10-cm plasticsquares
coatedwith Tanglefoot• that were hung in vegeta- the habitattypeswasgreaterin the fall than in
the spring.The greatestnumberof species(16)
tion at 0.5-m height intervals to 2.0 m [see Hutto
(1980) for discussionof the efficacyof this method]. wasrecordedduring both springand fall in the
creek-bottomhabitatand during the fall in the
At each site and in each season(1975 and 1976), 20
pine-fir forest.
boards (5 stationswith 4 boardseach, every 200 m
along the transectroute) were left hanging for 24 h
Within a given site in either season,an avbefore I countedthe insects.The boardswere hung
erageof 55%of all bird speciesof concernwas
in preciselythe samespotsin both seasons.The numwholly transient;that is, these specieswere
ber of insectscaptured per board is in itself an infoundin thesesitesonly during springor fall
sufficient estimate of prey availability to foliagemigration. This averageis conservativebecause
gleaners in different habitats, unless differencesin
an unknown proportion of individuals of some
vegetationdensity(foraging-substrate
availability)are
species were considered summer residents
taken into consideration.Assumingindependenceof
when, in fact, they were transient individuals
insects captured and vegetation density between
that bred farther north than the Chiricahua
habitats, I calculated a relative index of the number of
availableinsects/unitvolume of vegetation(ADJINS)
by multiplying the number of insectscaptured/site
timesa measureof vegetationdensity(VEGTOT/100)
Mountains. For example,one can be confident
that some(or most)of the Black-throated
Gray
Warblers(Dendroica
nigrescens)
sighted in the
within the site. The assumption that insect abundanceis independent of vegetation density is reason-
creekbottom during springwere transients,but
becausethe speciesis known to breed in that
able if different habitatsare involved. For example,
site, I (conservatively) called them all resia juniper woodland of the samevegetationdensity
dents. It probably is reasonableto assumethat
as an oak woodland would have many fewer insects
available to birds (Balda 1967). This assumptionalso 70-80%of the speciesor individualsin a given
is directlysupportedby a lackof correlationbetween site during either spring or fall were transient,
the unadjustedindex of insectabundance(as deter- but the only certainty is that the proportion
mined from sticky board samples) and either VEGONE (r = 0.23, NS) or VEGTOT (r = -0.35, NS).
exceeded
55%. The
bulk
of individuals
that
contributed to the patterns describedin this
January1985]
HabitatSelection
byMigratoryBirds
123
Verdin
Black-Tailed
Gnatcatcher
GrayVireo
MacGillivray's Warbler
Virginia's Warbler
Blue- Gray Gnatcatcher
Yellow
Warbler
Warbling Vireo
Nashville
Warbler
Bell'sVireo
Lucy's
Warbler
Bushtit
Solitary
Vireo
Golden-Crowned
Kinglet
Orange-Crowned
Warbler
Wilson's
Warbler
Townsend's Warbler
Hermit Warbler
Olive
Warbler
Grace's
Warbler
Hutton'sVireo
Black-Throated
GrayWarbler
Red-Faced
Warbler
PaintedRedstart
Ruby-Crowned
Kinglet
YellowRumped
Warbler
lOO
6'o
Similarif¾ Index
Fig. 2. Clusterdendrogrambasedon similaritiesin the seasonaldistributionsof 26 bird species.The 5
groupsof species
referredto in the text are indicatednumerically.
paper are, therefore, transient species, althoughthe analyses
includeresidentspecies
as
well.
There was a significant(ANOVA, seasonx
habitat;F = 2.7, P < 0.05)shift in the occupancy of the varioushabitattypesfrom springto
fall, as evidencedby the combinedbird densities;birds were relatively more abundantin
desertflatsand, especially,pine-fir forestin the
fall relative to the spring (Table 1). Moreover,
these changesin density were the result not
only of significantchangesin the densitiesof
bird speciesthatwere presentat a givenhabitat
during both seasons(69% of the bird species
exhibited significantchangesin density be-
separatelyto analyze their seasonalpatternsof
habitatuse,I identifiedfewer ecologicalgroups
that containedspecieswith similar habitat distributions in both seasons.This was done, first,
by looking for correlationsbetween the densities of all pairwise combinationsof species
using all habitats in both seasons.Sixty-four
(20%)of the possible325 pairwisecorrelations
were significant(r > 0.53, P < 0.05), and only
2 of thosewere negative.
I next identified groupsof specieswith similar patternsof habitatuseby transformingthe
correlation coefficientsinto similarity indices
(arccosinetransformation)and then subjecting
the similarityindicesto a clusteranalysis(av-
tween seasons),but also of some pronounced
seasonalshiftsin the kinds of speciesthat used
eragelinkage; Hartigan 1981).At the 60%level
of similarity,5 distinctspeciesgroupsemerged
eachhabitattype. On average,there was over (Fig. 2). Theseappearto be biologicallymean30% turnover (defined in Table 1) in species ingful groups rather than groups that were
from spring to fall in a given habitat. Consid- merely forced into existencethrough the clusering eachspeciesseparately,14 of 26 species tering procedure. The distribution and abun(54%) exhibited significant seasonalshifts in dancepatternsthat characterizeeachgroupare
habitatsoccupied(ANOVAs, seasonx habitat; outlined below.
P < 0.05).
Group 1.--There are 3 characteristicsshared
Ratherthan considereachof the 26 species by the 9 species
belongingto this group(Fig.
124
RICHARDL. HUTTO
[Auk, Vol. 102
January1985]
HabitatSelection
byMigratoryBirds
125
z•
126
RICn^RD
L. HUTTO
Increasing Elevation
[Auk,Vol. 102
16.4,P < 0.01).Thesespeciesshift markedlyto
the highest elevation forestsin the fall (ANOVA, season x habitat; F = 4.0, P < 0.01).
Group 4.--This is principallya mid-elevation
group (ANOVA, habitat effect; F = 44.1, P <
0.01) composedof 4 species(Fig. 3) whose densitiesdecreasesignificantly from spring to fall
(ANOVA, seasoneffect;F = 7.7,P < 0.01).They
becomesignificantly more broad in their habitat distribution from spring to fall (ANOVA,
0
20]
J-L_ Group
2
/ Fa,
/
Group
season x habitat; F -- 2.7, P < 0.05).
-•
Group 5.--Ruby-crowned Kinglet (Regulus
calendula)and Yellow-rumped Warbler (Dendroicacoronata)comprise this group (Fig. 3),
characterizedby mid- to high-elevation birds
o
40-
o••
(ANOVA,
•-,
50-
Group
n_
5J-L
n
0 --• Was-hes
I Pin•-Oak
I
Flats
Creek
Pine-Oak
Pi'ne J
Pine-Fir
Juniper
habitat effect; F = 6.3, P < 0.01)
whose densities decreasesignificantly from
spring to fall (ANOVA, seasoneffect;F = 17.2,
P < 0.01). They shift markedly from lower- to
higher-elevation forests from spring to fall
(ANOVA, season x habitat; F = 3.6, P < 0.01).
Thus,one canidentify groupsof speciesthat
Fig. 3. Frequencyhistogramsof the density of use the available range of habitatsduring the
birds in each of 7 habitat types during spring and spring and fall in a fashion that is similar to
fall seasons.
Bird speciesaresubdividedinto 5 groups, one another but differs from members of other
each representinga distinctive seasonalpattern of groups.In fact, one could use a higher level of
distribution
and abundance. The members of each
group are shownin Fig. 2.
similarity from the dendrogramto identify a
greaternumber of speciesgroups,eachdistributed
in distinct
fashion
within
and
between
seasons,but the characteristics
and biological
reality of suchgroupsbecomemore difficult to
3). They are basicallydesert species(ANOVA,
determine.
habitat effect; F = 18.8, P < 0.01), their densi-
Aspectsof vegetationstructure that differed
greatly among siteswithin a seasonwere canopy cover and mean vegetation height, while
between-season
changesin vegetationdensity
and insectdensity generally were pronounced
within a given site (Table 2). In general, the
densities of birds belonging to each of the 5
speciesgroupswere well correlatedwith at least
1 of the vegetation parameters(Table 3). The
desert groupswere negatively associatedwith
measuresof tall or densevegetation,while the
birds of higher-elevation habitats were positively associatedwith such measures.The index of insect abundance,with one exception,
was not stronglycorrelatedwith the densityof
any single group of birds, but it was significantly positively correlatedwith combinedinsectivorousbird density (Table 3).
Correlationsbetweenthe seasonalchangein
magnitudeof eachenvironmentalvariableand
the seasonalchangein bird density (Table 4)
show that, of the variablesmeasured,only the
ties increaseconsiderablyfrom spring to fall
(ANOVA, seasoneffect; F = 11.7, P < 0.01), and
they shift their habitat occupancytoward the
desert flats in the fall (ANOVA, season x habitat; F = 4.5, P < 0.01).
Group 2.--Bell's Vireo (Vireo bellii)and Lu-
cy's Warbler (Vermivoraluciae)comprisethis
group (Fig. 3), and their habitat distribution is
characterizedby restrictionto desert habitats
in both seasons (ANOVA, habitat effect; F =
16.5, P < 0.01). Their densitiesdecreasesignif-
icantlyfrom springto fall (ANOVA, seasoneffect;F = 8.2,P < 0.01),and they shift from desert flats and washesin the spring to exclusive
use of the desert-wash habitat in the fall (ANOVA, season x habitat; F = 4.7, P < 0.01).
Group 3.--This group is composed of 9
species(Fig. 3) that are basicallymid- to highelevation birds (ANOVA, habitat effect; F = 5.2,
P < 0.01)whosedensityincreasessignificantly
from spring to fall (ANOVA, seasoneffect;F =
January
1985]
Habitat
Selection
byMigratory
Birds
127
TABLE3. Pearson product-moment correlations between environmental variables and bird densities. Each
correlation
seasons.
was calculated
with
data from
the 69 censuses that were
conducted
in the 7 habitats
and 2
•
Groupb
1
2
3
4
5
All species
VEGTOT
-0.31'*
-0.39**
0.43**
0.59**
0.07
0.44**
VEGONE
0.55**
0.18
0.19
-0.23*
-0.39
ADJINS
CANCOV
-0.03
0.01
0.29*
0.56**
-0.05
0.18
PSRICH
HEIGHT
-0.53**
-0.39**
0.39**
0.44**
0.15
0.22*
0.16
-0.09
0.54**
-0.12
-0.56**
-0.41'*
0.45**
0.32**
0.24*
0.28
0.15
0.41'*
0.32**
* p < 0.05, ** P < 0.01.
Groups as defined in Fig. 2.
did change noticeablyfrom spring to fall, apparently in responseto the seasonalsummer
density of vegetation < 1 m high (r = 0.93, n =
7) and the index of insect abundance (r = 0.92,
n = 7) were able to account for the between-
rainfall
that
is characteristic
of southeastern
Arizona (Fig. 4). Some of the birds could have
been responding to the surge of vegetative
growth near the ground, which was striking in
the desertand pine-fir forest habitatsand was
seasonchange in bird density.
DISCUSSION
well
Habitatdistrbution
of migrants:
proximate
cues.The significantseasonalshifts in habitat distribution of mostmigratoryspeciesare of interest
becauseof the lack of visible changein many
cuesthat have been suggestedor implied to be
important settling cues by authors of habitatselection studies that were conductedduring
the breeding season(e.g. James 1971, Whitmore 1975).
correlated
with
the increase
in bird den-
sitiesin thosehabitatsfrom spring to fall (Table 4). Holmes et aL (1979) and Beedy (1981)
commented that low, understory vegetation
representsa unique foraging environmentthat,
when present,is capableof attractinga distinct
foraging guild.
Alternatively, the birds could have been responding to changesin food resourcesindependent of vegetation changes,which is possible because VEGONE and ADJINS are not
correlated (r = 0.37, NS), or to some combina-
Although someof the habitat variablespredict the density of birds over all habitats and
both seasonsin this study as well (CANCOV,
PSRICH, HEIGHT; Table 3), the same variables
tion of both variables.Baldaet al. (1975) argued
that shifts in food availability among habitats
do not changeseasonallywithin a habitat (Ta-
best accounted
ble 2). Therefore, these habitat variables are
cupancyof transientFlammulatedOwls (Otus
fiammeolus),although whether the authors
meant to imply a direct responseto food availability is uncertain. Austin (1970) also suspect-
unlikely to be the proximatecuesusedby the
birds for a settling response.The other three
habitat variables (VEGTOT, VEGONE, ADJINS)
for seasonal shifts in habitat oc-
TABLE
4. Pearsonproduct-momentcorrelationcoefficientsbetweenthe magnitudeof spring-to-fallchange
(factorby which springvalueis multipliedto give fall value)in bird densityand the magnitudeof springto-fall changein eachof the 6 environmentalvariables.aThere is a singlespring-to-fallcalculationfor each
variable in each habitat (n = 7).
Group•
I
2
3
4
5
All species
VEGTOT
0.42
0.89**
-0.35
-0.23
0.34
0.32
ß p < 0.05, ** P < 0.01.
Groupsas defined in Fig. 2.
VEGONE
ADJINS
CANCOV
0.90'*
0.24
0.26
0.03
0.72*
0.87'*
0.29
0.27
0.02
0.73*
- 0.24
-0.41
-0.18
0.14
-0.49
0.93**
0.92**
-0.45
PSRICH
0.65
-0.70*
-0.24
-0.06
0.38
0.47
HEIGHT
-0.52
0.22
-0.01
0.21
0.09
-0.48
128
RICHARD
L. HUTTO
[Auk,Vol.102
10
--J 6
._1
<•
fY 2
z
bJ
J
F
M
A
M
J
B.(r=o.o)
MONTH
o
Fig. 4. The yearly pattern of rainfall for southeasternArizona, as illustratedby 33 yr of datafrom
Portal, Arizona (Green and Sellers 1964).
.J
bJ
ed that changes in food availability could account for seasonalshifts in habitat use by the
bJ
.J
bJ
(r =1.0)
warblers of southern Nevada, but he added that
the temperature extremesin the lowlands during fall, rather than a decreasein food availability per se,could act to prevent use of such
areasby physiologicallyintolerant species.
Habitat distributionof migrants:ultimatefactors.--According to a recent school of thought,
as long as it does not limit bird populations,
food will be "loosely" exploited and will not
be an important factordeterminingbird species
presenceand abundance.This idea has led to
the developmentof the "checkerboard"model
(Rotenberryand Wiens 1980a,Wiens 1981),in
which an untilled checkerboardrepresentsan
unsaturated breeding habitat and a subsection
of the board representsone's study plot. From
year to year (or place to place within a year)
variations
in
the
kinds
and
abundances
of
speciesis very much a stochasticprocess,determined with about as much certaintyby the
compositionand abundanceof checkerswithin
a subset of the checkerboard
after the board has
been given a vigorous shake. Thus, according
to this model, one would not expect a close
correspondencebetween bird population densities and current food resourcelevels--they
would be independent (Fig. 5B). One would
expecta closecorrespondence
betweenbird and
food abundance only during an ecological
"crunch" (Weins 1977) year, when food resourcelevels are unusually low relative to bird
density (Fig. 5A). At the other extreme is the
possibilitythat, even though food may not be
limiting, competition still may exert a significant influence on bird community structure.
ß
cid.
\/
o
h
A
B
C
HABITAT
D
E
F
TYPE
Fiõ. 5. Graphical representationof the possible
relationships between the abundance of a predator
(histograms)and its prey (connecteddots) acrossa
seriesof habitat types.(A) The numbersof predators
may be limited by prey availability in each habitat,
in which casethere would be a closecorrespondence
betweenpredatorand prey densities.(B) Food may
not limit predator populations in most habitats, in
which casecompetition for food will not occur and
there will be a poor correspondencebetween pred-
ator and prey densities.(C) Foodmay not limit predator populations in most habitats, but competition
through the economicsof foraging will producea
close correspondencebetween predator and prey
densities.
This idea follows directly from optimal foraging theory (Krebs et al. 1983), which assumes
that competition is not an all-or-none phenomenon and that, although the strength of competition for food will vary with the degreeto
which it is limited, food is always limited to
someextentand the economics
of foragingwill
demand
that birds distribute
themselves
non-
January1985]
HabitatSelection
byMigratoryBirds
129
randomly in space.This resultsin a closematch
between food resourceproduction and bird
that affect the migratory birds while they are
abundance, no matter what the absolute level
A non-competition-based
hypothesisthat birds
of food availability(Fig. 5C).
The majority of bird species that passed
through southeasternArizona during migration was distributed significantlydifferently
over the availablehabitattypesin both spring
and fall. One can, however, identify groupsof
theseinsectivorousspecieswhoseseasonalpat-
farther north in summer or south in winter. (2)
settle in the best habitat (in terms of food avail-
ability) until some non-food-related resource
(e.g. space)becomesscarceand forces new arrivals to settle in the next best habitat (based
on food availability) until it too becomesmarginal (in terms of somenon-food resource),and
soon. On the basisof foodavailability,the suitterns of density and distribution are similar to ability of each habitat (site) would be identical
one another but very dissimilar to membersof for all birds in this case, but we would not exother such groups. Despite these differences pect a correlation between food levels and bird
amonggroups,when consideredtogether,the densities because the relative levels of the noncombined densities of the small, insectivorous food-relatedresources(suchasspace)would not
bird speciesmatch insect densities acrosshab- be expectedto be exactly the sameas the relaitats within a season (Table 3) and acrosssea- tive food levels amongsites.(3) The non-comsonswithin a habitat(Table4) remarkablywell. petition-basedhypothesisthat eachbird species
In other words,there are groupsof speciesthat is respondingto the presenceof a specificprey
respondmore or less independentlyof one type. The significantcorrelation between food
another, but when taken together provide a availability and bird abundance shown here
good fit to food resourceproduction. Such a may existonly becausethe diversity of prey is
correlationwould not be expectedaccordingto itselfcorrelatedwith my measureof foodavailnull models that assume independence be- ability. At this point I have no way to test
tween food availabilityand bird density.The whether a correlation exists between prey
distinction between the predictionsthat nec- speciesdiversity and my measureof food abunessarilyfollow from competition-based,and al- dance, but I can test another prediction that
ternative, null hypothesescannot be overem- necessarilyfollows from this hypothesis,i.e.
phasized.If foodresource-based
competitionis that bird speciesrichnessis correlatedwith my
absent,then one would not expecta correlation measureof food availability. This prediction
between food resourceavailability and num- follows becausea diversity of kinds of food
bers of consumers
over a variety of sites;in- would, accordingto this hypothesis,allow a
stead,all birdswouldbe expectedeitherto go diversity of bird speciesto occurin the habitat.
to thesamesite(i.e.wherefoodismostreadily Indeed, the correlation is significant (n = 14,
available)becauseuseof food by one individ- r = 0.61, P < 0.05), and such an explanation
ual would not decrease its availability to seemsplausible on this basis.It remains for fuanother,or to distributethemselvesrandomly ture researchto determine whether high seawith respectto food availability(null model). sonal turnover in insect speciesunderlies the
It is important to appreciatethat a null model high seasonalturnover in bird specieswithin
does not predict, for example, 10 times more a site, as would also be predicted by this hybirds in a site that has 10 times more food than
pothesis.(4) The competition-based
hypothesis
anothersite;that would be the caseonly if the that "crunch" conditions existed during each
area were 10 times larger.
year and seasonthat I was present in Arizona.
Alternative hypothesesthat do generatepre- This is unlikely, however, becauseeach of the
dictionsconsistentwith my resultsinclude the years was normal in terms of rainfall (Cody
following: (1) A non-competition-based
hy- 1981:Table 1). Finally, (5) the competition-based
pothesisthat the populationsof both predator hypothesisthat food availability determines,in
(birds) and prey (arthropods) were affected an ultimate sense, the distribution and abunsimilarly,but independently,by the samemor- danceof the insectivorousbird speciesduring
tality factors (e.g. weather). This is most un- migratory passage.Becausethe combination of
likely becausethe migratorybirds are present all bird groupsprovided the bestfit to resource
for a matter of days,and the physicalcondi- availability, this hypothesiswould necessitate
tions that affectarthropodpopulationsin Arithe presenceof interspecific,as well as intrazona undoubtedly are not the sameconditions specific,competition to explain the organiza-
130
RICHARD
L. HUTTO
tion of bird communities
[Auk,VoL 102
at these times of the
highly predictableand the migrantshave been
"programmed" to shift their habitat use acBy what mechanismwould birds be able to cordingly.In either case,food availabilitycould
achieve close matches to food resource levels
be of ultimate importancein producing such a
through competitive adjustments,as would be closecorrespondencebetween the distribution
necessitatedby the fifth hypothesisabove?Ac- and abundanceof birds and their prey.
cording to the "crunch" hypothesis (Wiens
Theseresultssuggestthe possibilitythat sea1977), under most environmental conditions it
son-to-season
or year-to-yeardifferencesin bird
would be rare, if not impossible,for the kinds speciescompositionand densities on a given
and densitiesof speciesto be able to match re- plot (evenbreeding-season
studyplots)maybe
sourceavailability at any particular point in largely the result of nonrandom food assesstime becausethe individuals present at that ment processessuch as facultative settlement
point in time are the survivors or descendants and migration (Serventy 1971, Ward 1971, Fretof survivors that were able to "squeeze" well 1972, Sinclair 1978, Pulliam and Parker
through very different environmental condi- 1979, Wiens and Rotenberry 1981, Smith 1982)
tions at some time in the past (Wiens 1977, Em- or "flyover" mechanisms(Nudds 1983).By this
len 1981, Emlen and DeJong 1981). However, I do not mean to suggestthat food-basedcomthis view assumesthat food levelsvary unpre- petitive interactionsare equally intense at all
dictably and that "... low-fecundity taxa such times and places. Rather, becausecompetition
asmany birds and mammalsmay be quite lim- is not an all-or-none phenomenon,the level of
ited in their capacityto generate population competitive interactions that does exist (megrowth sufficient to track variation in food re- diated either directly or indirectly through relsourcelevels at all closely" (Wiens 1977: 593). ative food abundancesamong or within sites)
Whether or not the statisticallysignificantbut may be responsiblefor much of the changethat
otherwise weak matches between food levels
occursin community structure from one place
and bird densities demonstrated
herein are
or time to another. More definitive explanacloserthan expectedon the basisof the crunch tions for the observedshifts in habitat useby
hypothesisis not really clear. Nonetheless,the thesemigratory birds must await testing of adhypothesisplays down the importanceof flex- ditional predictionsthat follow from the plauibility in foragingbehavior (Alatalo 1980,Hut- sible alternativehypothesespresentedabove.
to 1981),which might enable birds to shift their
habitat occupancyor densitieswithin habitats
ACKNOWLEDGMENTS
year.
to match
current
relative
food-resource
condi-
tions, such as some woodpecker speciesseem
to do within a year following a fire (Blackford
1955, Bockand Lynch 1970, Theberge 1976) or
ducksmay do from year to year (Nudds 1983).
Terrill and Ohmart (1984) also describe a facul-
tative migratory movement pattern of the Yellow-rumpedWarbler that apparentlyis related
to food availability.
The habitat shifts reported herein that correspondwith shiftsin food availabilitymay resuit from birds being able to "test" severallocations before settling in any one for their
typical 4-6-day (unpubl. banding records)
stopoverperiod. Suchresourcetracking may be
somewhatuniquein that adjustments
that "finetune" bird community compositionto resource
production may be much easier to accomplish
during migration than during other times of
the year. It also is possiblethat, although the
relative food levels amonghabitatschangedramatically between seasons,such change is
Field work during springand fall is academically
awkward and would have been impossiblewithout
the financial support provided by the Frank M.
Chapman Fund of the American Museum of Natural
History and the ResearchAdvisory Council of the
University of Montana. To theseorganizationsI am
grateful.I alsowishto thankMyra ShulmanandMary
Harris for their assistance in the field and Kimball
Garrettand Martin Cody for accessto their unpublished breeding bird censusdata. The manuscript
benefitedfrom discussions
with, and reviewsby, S.
J. Wright and J. R. McAuliffe and from comments
providedby O. J•rvinenand anonymousreviewers.
LITERATURE CITED
ALATALO,
R.V. 1980. Seasonaldynamicsof resource
partitioning among foliage-gleaning passetines
in northern Finland. Oecologia45: 190-196.
AMERICAN ORNITHOLOGISTS'UNION.
1983. Check-list
of North American birds, 5th ed. Baltimore,
Amer.
Ornithol.
Union.
AMMAN,G. D., & P. H. BALDWIN,1960. A compari-
January
1985]
Habitat
Selection
byMigratory
Birds
son of methodsfor censusingwoodpeckersin
spruce-firforestsof Colorado.Ecology41: 699-
131
HOLMES, R. T., R. E. BONNEY, & S. W. PACALA. 1979.
Guild structure
706.
of the Hubbard
Brook bird com-
munity:a multivariateapproach.Ecology60:512520.
ANDERSON,B. W., & g. D. OHMART. 1977. Climato-
logicaland physicalcharacteristics
affectingavi- HUTTO, R. L. 1980. Winter habitat distribution of
migratory land birds in western Mexico, with
an population estimatesin southwesternriparispecialreferenceto small, foliage-gleaninginan communitiesusing transectcounts.Pp. 193sectivores.Pp. 181-203 in Migrant birds in the
200 in Importance, preservationand manageNeotropics:ecology,behavior,distribution,and
ment of riparian habitat: a symposium(R. R.
Johnsonand D. A. Jones,technicalcoordinators).
U.S. ForestServ. Gen. Technical Rept. RM-43.
AUSTIN,G.t. 1970. Migration of warblers in southern Nevada. Southwestern Naturalist 15: 231-237.
BALDA,R. P. 1967. Ecologicalrelationshipsof the
breedingbirdsof the ChiricahuaMountains,Ar-
izona.UnpublishedPh.D. dissertation.Urbana,
Univ.
MCKNIGHT, & C. D. JOHNSON. 1975.
Flammulated Owl migration in the southwestern United
BEEDY,E. C.
biers. Auk 98: 765-777.
JAMESß
F. C. 1971. Ordinations of habitat relation-
shipsamongbreedingbirds.WilsonBull. 83:215236.
Illinois.
, B.C.
--.
conservation (A. Keast and E. S. Mortonß Eds.).
Washington, D.C.ßSmithsonian Inst. Press.
1981. Seasonalvariation in the foragingbehavior of some migratory western wood war-
States. Wilson Bull. 87: 520-533.
1981. Bird communities and forest
structure in the Sierra
Condor 83: 97-105.
Nevada
of California.
KREBS,J. R., D. W. STEPHENS,
& W. J. SUTHERLAND.
1983. Perspectivesin optimal foraging.Pp. 165216 in Perspectives
in ornithology(Alan H. Brush
and G. A. Clark, Jr.,Eds.).New York, Cambridge
Univ.
Press.
MACARTHUR,
R. Hß 1969. Speciespackingand what
BLACKFORD,
J. L. 1955. Woodpeckerconcentration
interspeciescompetition minimizes. Proc. Natl.
in burned
forest. Condor
57: 28-30.
BOCK,C. E., & J. F. LYNCHß1970. Breedingbird pop-
Acad. Sci. 64: 1369-1371.
ulations of burned and unburned conifer forest
1970. Species packing and competitive
equilibrium for many species.Theoret. Popula-
in the Sierra
tion Biol. 1:
Nevada.
Condor
72: 182-189.
Cot•¾,M. L. 1974. Competitionand the structureof
bird communities.Monogr. in Population Biol.
No. 7, Princeton, New Jersey, Princeton Univ.
Press.
ß 1981.
Habitat
selection in birds: the roles of
1972. Geographicalecology:patternsin the
distribution of species.New York, Harper and
Row.
Nvl•l•S, T. D. 1983. Niche dynamics and organization of waterfowl guilds in variable environ-
ments.Ecology64: 319-330.
vegetationstructure,competitors,and producPULLIAM,H. R., & T. H. PARKER.1979. Population
tivity. BioScience31: 107-113.
regulation of sparrows.Fortschr.Zool. 25: 137EMLEN,
J.t. 1978. Densityanomaliesand regulatory
147.
mechanismsin land bird populations of the
ROBINETTE,
W. L., C. M. LOVELESS,
& D. A. JONES. 1974.
Florida peninsula.Amer. Natur. 112: 265-286.
Field tests of strip census methods. J. Wildl.
--.
1979. Landbird densitieson BajaCalifornia
Islands. Auk 96: 152-167.
Mgmt. 38: 81-96.
J. t., & J. A. WIENS. 1980a. Temporal
ß 1981. Divergencein the foraging responses ROTENBERRY,
variation in habitat structure and shrubsteppe
of birds on two Bahamaislands.Ecology62:289295.
bird dynamics.Oecologia47: 1-9.
ß & M. J. DEJONG. 1981. Intrinsic factors in ---,
&
. 1980b. Habitat structureßpatchiness, and avian communities in North American
the selectionof foragingsubstratesby Pine Warbiers: a test of an hypothesis.Auk 98: 294-298.
steppevegetation:a multivariate analysis.EcolFRANZRIBßK. E. 1981. The determination of avian
ogy 61: 1228-1250.
densitiesusingthe variable-stripand fixed-width SERVENTY,
D. L. 1971. Biology of desert birds. Pp.
287-339 in Avian biology, vol. 1 (D. S. Farner
transectsurveying methods.Pp. 139-145 in Esand J. R. King, Eds.). New York, Academic Press.
timating the number of terrestrialbirds (C. J.
Ralph and J. M. Scott,Eds.).StudiesAvian Biol. SINCLAIR,
A. R. E. 1978. Factorsaffecting the food
No. 6.
supplyand breedingseasonof residentbirdsand
FRETWELL,
S.W. 1972. Populationsin a seasonalenmovementsof Palaearcticmigrants in a tropical
African savannah. Ibis 120: 480-497.
vironment. Monogr. in Population Biol. No. 5,
SMITHß
K.G. 1982. Drought-inducedchangesin aviPrinceton, New JerseyßPrinceton Univ. Press.
GREEN,C. R., & W. D. SELLERS
(Eds.). 1964. Arizona
an community structure along a montane sere.
climate. Tucsonß Univ. Arizona Press.
Ecology63: 952-961.
HARTIGAN,
J. 1981. Clusteranalysisof variables.Pp. TERRILLßS. B., & g. D. OHMART. 1984. Facultative
448-455 in BMDP statistical software 1981 (W. J.
extensionof fall migration by Yellow-rumped
Warblers (Dendroicacoronata).Auk 101: 427-438.
Dixon, Ed.). Berkeley,Univ. California Press.
132
RICHARDL. HUTTO
THEBERGE,
J.B. 1976. Bird populationsin the Kluane
Mountainsßsouthwest Yukonßwith special ref-
erenceto vegetationand fire. CanadianJ. Zool.
54: 1346-1356.
TILGHMAN,N. G., •r D. H. RUSCH. 1981. Comparison
of line-transectmethodsfor estimatingbreeding
bird densities in deciduous woodlots. Pp. 202-
208 in Estimatingthe numbersof terrestrialbirds
(C. J. Ralph and J. M. Scott,Eds.).StudiesAvian
Biol. No. 6.
WARD,P. 1971. The migratior•patternsof Quelea
queleain Africa. Ibis 113:275-297.
WHITMORE,R. C. 1975. Habitat ordination of passerine birds of the Virgin River Valleyß southwestern
Utah. Wilson
Bull. 87: 65-74.
From "General Notes" (1885 Auk 2: 113-114):
[Auk, Vol. 102
WIENS,J.A. 1977. On competitionand variable environments.
Amer.
Scientist
65: 590-597.
1981. Scale problems in avian censusing.
Pp. 513-521 in Estimatingthe numbersof terrestrial birds (C. J.Ralph and J.M. ScottßEds.).Studies Avian
Biol. No. 6.
ß 1983. Avian community ecology:an iconoclasticview. Pp. 335-403 in Perspectivesin ornithology (Alan H. Brush and G. A. Clark, Jr.,
Eds.). Cambridge, England, Cambridge Univ.
Press.
ß & J. T. ROTENBERRY.1981. Habitat associa-
tionsand communitystructureof birdsin shrubsteppeenvironments.Ecol. Monogr. 51: 21-41.
familiar in the case of the horse whose back or shoul-
der is galled by the harness;white patchesappear,
owing to loweredvitality of the injuredpart. These
"Albinism.--My attention was drawn to a note in
the 'O61ogist'for April last,in which the writer gives casesare familiarßbut I wish to give possiblyanother
his experiencein albinismand asksfor an explana- causeactingin the sameway, only moregeneral.It
tion of these freaks of nature. In order to air my exis this.When a boy I shotamongothersa blacksquirit havinga perfectlywhite tail,
perience,and at the sametime to give a probable rel peculiarlymarkedß
causeßwhich I would like, for the sake of possible with some white about the head; on making a post
verificationß other observers to look for in the futureß
morteraI discoveredthrough a rent in the intestines
a tape-wormabout20 feet in length. Did not wonder
is the objectof the present note.
then that his head was gray. A few yearsafter a par"True albinism is of course congenital, and is a
condition in which the normal pigmentary matter is tially white Red-wingedBlackbird(Agelaeus
phoenideficientin the systemof the individualaffected;in ceus)was takenßwhich also contained two or three
suchcasesthe eyesare pinkßand the skin with its taenia; next a partial albino Mallard; then a Robin
with a white head and mottled
appendages
are white or nearly so. In the caseof (Turdusmigratorius)
partialalbinos,however,it is difficult;their condition can probably be explained by some circumstancesoccurringafter birth which will accountfor
the changein the colorof the skinßsuchfor instance
as the casegiven by the writer in the 'O61ogist,'in
which the skin had been injured on the back of a
Swift, and next year the patch of white feathersindicated the situation of the injury. The samething is
back and breast. All were mountedß and are now in
my collection.Eachof thesehad two or more tapeworms in their intestines. I am aware that birdsß es-
peciallysomespecies,are particularlyobnoxiousto
tape-worms,and the abovemay have been merely
coincidences;still it has been observed sufficiently
often to make the fact suspiciousas a causeof albinism.--G. A. M'CALLUM, Dunnville, Ont."