PLUMAGE, MORPHOMETRIC, AND SONG VARIATION
MOURNING (OPORORNIS PHILADELPHIA) AND
MACGILLIVRAY'S (O. TOLMIEI) WARBLERS
IN
JAY PITOCCHELLI
Ornithology
Department,
American
Museumof NaturalHistory,CentralParkWestat 79thStreet,
New York,New York 10024USA,andBiologyDepartment,
Queens
College,
Flushing,
New York11367USA
Al•sTP•CT.--Mourning(Oporornis
philadelphia)
and MacGillivray's(O. tolmiei)warblersare
currently recognizedasdistinctspecies(AOU 1983).The specificstatusof thesetaxa,however,
hasbeenquestionedbasedon morphologicalsimilarityand reportsof hybridizationin central
Alberta (Cox 1973). I investigated the distinctnessof these taxa by comparing plumage,
skeletal,and primary songcharactersfrom fresh collectionsand recenttaperecordingsfrom
the allopatricportionsof their breeding rangesand the potential contactareas.The plumage
analysesrevealed overlap in charactersoriginally usedto diagnosethesetaxa;however, the
incidence of extreme specimenswas low. Separatemultivariate analysesof morphological
and songcharactersshowedthat the taxaare essentiallydistinct in multivariatespacewith
little or no overlap.The Mourning Warbler was larger for mostskeletalcharactersand had
lower songfrequencies.Plumage charactersoriginally used to diagnosethese taxa separate
a majority of specimens,but they are ineffective for distinguishingextreme variants or
hybrids.Songtype proved 100%reliable in discriminatingbetweentaxa.I found no hybrid
contactbetween thesetaxa.Furthermore, specimenscollectednear the potential contactzones
were as variable morphologicallyas specimensfrom the allopatric portionsof the breeding
ranges.I believe thesetaxashouldcontinueto be considereddistinct species.Received
10May
1989,accepted
9 September
1989.
MOURNING (Oporornis philadelphia) and
MacGillivray's (O. tolrniei)warblersform an eastwest speciescomplex, whose breeding ranges
meet in the northern Great Plains (sensuRising
1983).The Mourning Warbler is the easterntax-
imens.Further examinationof Mourning Warbler specimensfrom different partsof its breeding range revealed the presenceof intermediate
specimensthat possessed
either dark lores,eyearcs, absence of black bibs, or some combination
on; it breeds in boreal forest from Newfound-
of these (Chapman 1917, Hall 1979). Lanyon
and Bull (1967)acknowledgedthe equivocalnato West Virginia, partsof Michigan, Wisconsin, ture of plumagecharacters
and usedan external
Minnesota, and North Dakota. MacGillivray's measurement(Wing minus Tail: W - T) to sepWarbler breeds in riparian habitat and dis- arate98%of Mourning and MacGillivray'swarturbed secondgrowth in the RockyMountains bler specimens.Their results, however, were
from northern Arizona to Alaska. Both taxa were
basedexclusivelyon samplesfrom the allopatdescribedas separatespeciesby Baird (1858) ric portions of the breeding ranges.Kowalski
land to northeastern
British Columbia
and south
and are still considered separate (AOU 1983).
(1983) found much more overlap in W - T, us-
Difficulty in identifying intermediate speci- ing specimensnear the potential contactzones.
mens,however, hascausedsomeornithologists This result suggestshybrid contactbetween the
(Chapman i917, Phillips 1947, Hofslund 1962,
Mengel 1964,Mayr and Short 1970)to speculate
that these taxa are easternand western subspecies.Accountsof hybrid contact(Cox 1973,Salt
1973) have castdoubt on their specificstatus.
The doubtsare basedon the equivocalnature
of plumage charactersoriginally used to diagnose these taxa (Table 1). Characters used in
diagnosesare presumedto be unique to a given
taxonand provide 100%discriminationof spec-
taxa.
Cox (1973) collected some Mourning Warbiers with "MacGillivray's-like" charactersin
centralAlberta and concludedthat the taxahybridized there. Taverner (1919) collected a sus-
pected mixed pair (Mourning male and
MacGillivray's female) in Nevis, Alberta; but,
based on the W - T measurement, this female
falls into the range of the Mourning Warbler
(Hall 1979). Hall (1979) was also not convinced
161
The Auk 107:161-171.January1990
162
JAYPITOCCHELLI
[Auk,Vol. 107
TABLE
1. Plumagecharacters
usedto diagnoseadultMourningand MacGillivray'swarblers.
Character
MourningWarbler
MacGillivray'sWarbler
Lores a
Absent
Eye-arca
Absent in adults, present in immatures
Dark in males
Present in adults and immatures
Black bib •
Concentration
Diffuseblackfeatheringon throat
of black feathers on
lower throat and upper breast,
present in malesonly
>- 10 mm
Wing-minus-tail b measure
and upper breastof males,
absent in females
-<11 mm
Baird 1858, Coues 1903, Hall 1979.
Phillips 1947,Lanyon and Bull 1967.
that all intermediatespecimensreportedby Cox
and others (Patti and Meyers 1976, Beimborn
1977) were hybrids because he found intermediatespecimensof Mourning Warblersfrom
the allopatric portions of its breeding range.
These
extreme
variants
from
eastern
Canada
raise the question whether the intermediate na-
ture of thesespecimensis due to hybridization
or falls within the normal range of variability
of either
taxon.
My goal was to clarify the specificlimits of
these warbler
taxa based on the distinctness
of
eachtaxon.I concentratedmy analysison males
for two reasons:intermediate males show "hybrid
characters"
better
than
females
and their
singing behavior can be recorded and compared. I collected and compared fresh speci-
lected. Recordingswere made with a Uher-4000 Report Stereoand Dan GibsonE. P.M. 300 microphone.
At least 10 songsper male were recordedbeforecollection. At most localities,>-30 songswere recorded
from at least one male.
Morphology.--Because
of the closemorphological
resemblance
of thesetaxaand the paucityof plumage
charactersuseful for diagnosis,hybrid indices were
not consideredappropriate.I scoredall study skins
for the presence
or absence
of eye-arcs
anddarklores.
Flattenedwing measurements
were madewith a wing
ruler (nearest 0.1 mm) and tail measurements were
madewith Max-cal Calipers(nearest0.1 mm) according to Lanyon and Bull (1967). I also counted the
number of males with W - T measurements that fell
within the intervals of (W - T) < 8 mm, 8 mm <
(W-T)
<9mm, 9mm<(W-T)<
10mm, 10mm
< (W-T)
< 11 ram, 11 mm< (W-T)
< 12ram,
12 mm< (W-T)
< 13 ram, (W-T)
> 13 min. I
mensfrom the allopatricportionsof the breed- usedhistogramsto displaythe frequencyof Mourning ranges and from the hypothetical contact ing vs. MacGillivray's warbler specimenswith these
areas,and ! evaluated the range of variability characters.
The songtype of eachspecimenwasalso
and overlap in traditional plumage characters noted.
Analysesof skeletal charactersprovide an indeused to diagnosethese taxa. I added two new
suites of skeletal
and behavioral
characters
to
pendent test of resultsobtained from plumage anal-
the analysis,and determined if specimensof
thesetaxa occupydifferent "morphologicalor
song space" based on principal components
analyses(PCA)of morphologicalandsongcharacters.Finally, I evaluatedsongsyllablesharing
yses(Troy 1985).Morphometricanalysesof skeletal
charactershave been used successfullyto discriminate sibling taxa of meadowlarks(Sturnella;Rowher
1972)and wood-pewees(Contopus;
Risingand Schue-
by these taxa.
(BDEP),nasalbone width (NASW), interorbital width
Fieldmethods.--Adult males were collected during
the breedingseason,between6 Juneand 15 Julyfrom
1983to 1986(Fig.1).In 19851concentrated
my efforts
in the potential contact areas in western Canada.
Specimenswere prepared in the field as flat skinsand
are at the American
andSchnell(1971):premaxilla
length(PRL),bill depth
(INORW), skull width (SKW), skull length (SKL),
mandiblelength(MANL), mandibledepth(MAND),
MATERIALS AND METHODS
skeletons. All materials
ler 1980).I used25 skeletal dimensionsfrom Robbins
Museum
of Natural History.
Whenever possible,I madetape recordingsof Oporornissongsbeforecollection.Not all specimenswere
recorded, however, nor were all recorded birds col-
coracold length (CORL), scapula width (SCAPW),
sternum length (STERL), keel length (KEEL), keel
depth(KEED),minimumsynsacrum
width (SYNMW),
maximumsynsacrumwidth (SYNW), femur distalend
width (FEDW),femurlength(FEL),tibiotarsus
length
(TIBL), tarsometatarsus
length (TARL), tarsometatarsusdepth(TARD),humerustrochanterlength(HTRL),
humerusdistalend width (HDEW),humeruslength
(HUML), ulnalength(ULNL),carpometacarpus
length
(CARPL). Skeletal dimensions were measured with
Max-cal Calipers to the nearest0.1 mm and entered
January
1990]
Oporornis
Systematics
163
directly into a NEC Portable Computer (PC8201A)
using Lessoft(version 1.0, Marcus 1982).
Song.--Songrecordingswere analyzed on a Kay
Elemetrics6061 Sona-Graphusing wide band filter.
Terminologyof componentsand physicalparameters
follows Shiovitz (1975) and Baptista(1977). A song
noteis any continuoussoundtracingon a sonograph.
A syllable
is a collectionof notes,and a songis a collection of syllables.The different parts of primary
song(I, II, III) in thesetaxacontaineda uniquesingle
syllabletyperepeatedseveraltimes.Songswereeither
monosyllabic
(partI only), disyllabic
(partsI and II), or
trisyllabic
(partsI, II, III).
Speciesdifferencesin primary songmostoften occur in syllable/notemorphologyand/or frequency
parameters(Becker1982).I visually inspectedsyllables from each song and catalogedthem basedon
differencesin grossmorphology(syllablecatalogsin
Pitocchelli1988).I then comparedsyllablesfrom both
catalogsto determinethe amountof syllablesharing
between
these taxa.
I analyzed differencesbetween these taxa in the
numberof partsper songbasedon differentsyllable
types(NPSO),numberof syllablesper song(NSSO),
duration of song (DUR), minimum song frequency
(MINS), maximumsongfrequency(MAXS), number
of notesof the first syllablefrom part I of the song
(NNA), numberof notesof the firstsyllablefrompart
II of the song(NFA), maximumfrequencyof the first
syllablefrom part II of the song(MAFA), and maximum frequencyof the secondsyllablefrom part II of
the song (MAFB). Sonographswere measuredwith
Max-calCalipersto the nearest0.1 mm and entered
into a NEC Portable Computer-PC8201A (Marcus
1982). These measurementswere later converted into
kilohertz and seconds.I paid specialattention to the
songsof intermediatespecimens(basedon plumage
or skeletal materials) of both taxa.
Statistics.--Multivariate
analysesreorganizethe total variation among correlatedvariablesto a new set
of uncorrelatedvariables.Several multivariate approacheshave been applied in pheneticanalysesinvolving the discriminationof operationaltaxonomic
units (OTUs) (Sokaland Sneath1963,Thorpe 1976).
I used principal componentsanalysis(PCA) to investigatethe distinctnessof thesetaxain multivariate
spacebasedon external study skin measurements,
skeletalmeasurements,
and physicalparametersof
song. The following variables were used in these
analyses:external measurements--Wing,Tail, Wing
minus Tail (W - T); skeletal measurements--NASW,
MANL, CORL, KEEL,FEL, ULNL, HUML; and song
parameters--NPSO,NSSO,DUR, MINS, MAXS, NNA,
NFA, MAFA, MAFB. Analysesof thesedatasetswere
conductedusing PROC PRINCOMP in SAS (version
5.16, 1985).Raw datawere logso-transformed
prior to
each multivariate analysis.I extractedPC scoresfor
each specimenalong the first three principal componentaxesfrom a variance-covariance
matrix. I plot-
Fig. 1.
Sampling localities of Mourning and
MacGillivray'swarblers(preciselocalitiesare available from the author).
ted the principal component(PRIN) scoresin three
dimensional space along the PRIN1, PRIN2, and
PRIN3 axes.I rotatedthe plotsalong the x, y, and z
axesusing MACSPIN (version 2.0, 1988) until maximum separationof Mourning and MacGillivray's
OTUs
was achieved.
RESULTS
Studies of hybridization in birds have includedsamplesfrom pure populationsfor comparisonswith specimensfrom the contactzones
(Rising 1983). I made collectionsand tape recordingsof Mourning and MacGillivray'swarbiers from pure, well-marked populations in
the allopatric regions of their breeding ranges
and from the potential contactareasin Alberta
and BritishColumbia.Mourning Warblersfrom
Ontario, New York, and Quebec represented
the allopatric portion of their breeding range.
Allopatric samplesof MacGillivray's Warblers
came from south-central
British
Columbia,
Washington,Oregon, California, Idaho, Nevada, Montana, and Wyoming. I collected 534
males of both taxa for plumage and skeletal
analyses.The allopatric samplescontained218
Mourning and 236 MacGillivray's warblers. In
1985, I sampled contact areas previously reported by Erskine and Davidson (1976) in
northeastern British Columbia, and by Cox
(1973) and Salt (1973) in central Alberta. I col-
lected and recorded birds along the Alaska
Highway from the Liard River Hot Springsto
Fort Nelson
in northern
British
Columbia.
In
Alberta, I sampled from Lesser Slave Lake
through Whitecourt, Obed, BrazeauReservoir,
Battle Lake, Red Deer to the Sheep River. This
transectalsoincludedthe sitesstudiedby Cox
164
JAYPITOCCHELLI
25O
A
MacGillivray's Warbler
225
200
[Auk, Vol. 107
I taped singing males from the same localities. I recorded 137 Mourning Warblers from
eastern Canada and 58 males from the contact
areas.MacGillivray'smales(116)were recorded
150
in the northwestern
• 25
British Columbia.
• 00
United
States and southern
I recorded
19 males near the
potential contactareas.
50
25
MORPHOLOGICAL CHARACTERS
o
25o
225
Externalcharacters.--Therewas some overlap
Mourning
Warbler
in plumage characters(Fig. 2). MacGillivray's
Warblerswere not variable for dark loresor eyearcs. Extreme MacGillivray's Warblers resembled Mourning Warblers only in the wing-
2o0
175
•50
125
100
minus-tail
75
25
0
Lores Lores
Dark
Absent
B
Eye-arcs
Ill
I.I.
60
i.u
40
bia. Four of these birds were from the allopatric
sangonly MacGillivray'ssong.Plumagecharactersof Mourning Warblers were more vari-
able (Fig. 2). Thesebirds either had dark lores,
eye-arcs,
(W - T) < 11mm,lackeda blackbib,
Ill
1•
tial contact areas in Alberta and British Colum-
the contact areas. All these extreme specimens
eo
o
Extreme
regions(southernBritish Columbia, Washington, and Oregon) whereastwo camefrom near
u'• 1oo
-
(W - T) measurement.
MacGillivray'sWarblerswere found in the allopattic partsof the breedingrangeand poten-
50
or some combination
20
of these characters.
Al-
though most Mourning Warblers lack dark
z
lores, males with lores were found in small
60-
Mourning Warbler
50-
'
•0-
20-
.
10-
numbers throughout the breeding range. Extremebirdswith eye-arcswere alsouncommon
(Fig. 2). There was no segregationor increase
in intermediatespecimensnear the contactareas.
For instance,there were higher incidencesof
maleswith eye-arcsin someOntariolocalities
(Dorion:11.1%maleswith eye-arcs;Geraldton:
14.2%;and Cochrane: 16.6%) than in the potential contact areas in central Alberta (5.2%). All
extreme Mourning Warbler specimens sang
Mourning Warbler songs (Fig. 3: A, C). Al-
thoughtheseexternalcharacters
will work for
mostspecimens,
they are equivocalfor extreme
Fig. 2. Histogramsof MacGillivray'sand Mourning warblerspecimens
with traditionalcharacters
used
to diagnosethesetaxa.(A) Frequencydistributionsof
eye-arcsand dark lores. (B) Frequencydistributions
of specimenswith wing-minus-tail (W - T) measurements (mm) falling in specifiedintervals.
specimens.
MacGillivray's
Warblersaveragedlongertails
than Mourning Warblers,which accountsfor
the smaller W-
T measurements (Table 2).
Principalcomponents
analysesof externalmeasurements(W, T, W - T) revealed little overlap
in multivariate space(Fig. 4: A). Wing-minus(RockyMountainHouse,Caroline,Kananaskis) tail measurementhad the highest loading on
and Salt (Pigeon Lake) in 1973. I collected49 PRIN1 and contributed most to the separation
MourningWarblersand 29 MacGillivray'sWar- alongthe PRIN1 axis(Table3). PRIN1 accountbiers from these potential contact areas.
ed for 99% of the variance. In most morpho-
January
1990]
Oporornis
Systematics
165
6
kHz
4
2
\8/\\
6
kHz
4
2
6
kHz
4
2
6
kHz
4
2
0.5
1.0
Seconds
Fig. 3. Sohogramsof Mourning and MacGillivray'swarbler songs.(A) Ontario Mourning Warbler male
(AMNH 13365)with dark lores;(B) Maine Mourning Warbler male (AMNH 13263)lacking eye-arcsor dark
lores;(C) QuebecMourning Warbler male (AMNH 13333)with eye-arcs;(D) MacGillivray'sWarbler male
(AMNH 14358);(E) MacGillivray's
male(AMNH 14362);(F) MacGillivray's
male(AMNH 14363).The three
MacGillivray'smales(D-F) were neighborsfrom Jarbidge,Nevada.AMNH refersto AmericanMuseumof
NaturalHistoryskeletalspecimencatalognumbers.Warblersketchesby Ken Davignonand JayPitocchelli.
166
JAYPITOCCHELLI
TABLE2. Elementarystatisticsof Oporornis
skeletal
and external
characters.
[Auk, Vol. 107
TABLE
3. Eigenvectors
of principal componentanalyses(PCA) of external,skeletal,and songcharacters. a
MacGillivray's
Charac-Mourning
Warbler
ter a
PRL
BDEP
NASW
INORW
SKW
SKL
MANL
MAND
CORL
SCAPW
STERL
KEEL
KEED
SYNMW
SYNW
FEDW
FEL
TIBL
TARL
TARD
HTRL
HDEW
HUML
ULNL
CARPL
WEIGHT
WING
TAIL
W - T
n
Mean (SD)
230 13.72 (0.62)
245
1.12 (0.11)
249
5.95 (0.33)
240
3.03 (0.24)
221 13.35 (0.31)
194 30.74 (0.73)
235 22.32 (0.62)
252
1.51 (0.08)
238 14.59 (0.36)
251
1.94 (0.18)
222 16.44 (0.52)
222 14.71 (0.60)
225
6.32 (0.34)
220
8.90 (0.27)
235
5.04 (0.26)
242
2.61 (0.08)
232 15.48 (0.44)
156 28.14 (0.71)
189 20.57 (0.62)
229
2.21 (0.23)
249
3.64 (0.15)
244
3.30 (0.14)
236 14.50 (0.30)
187 16.53 (0.50)
159
9.46 (0.36)
239 12.27 (0.67)
270 61.56 (1.95)
263 49.32 (2.01)
261 12.21 (2.00)
Warbler
n
176
180
170
165
121
122
167
187
184
191
165
171
172
152
168
188
171
44
146
174
164
151
136
129
135
139
211
202
202
Character
External
Mean (SD)
13.43 (0.54)
1.06 (0.12)
5.62 (0.28)
2.66 (0.18)
13.38 (0.25)
30.42 (0.60)
21.89 (0.54)
1.39 (0.08)
13.81 (0.37)
1.78 (0.14)
15.37 (0.54)
13.58 (0.66)
5.84 (0.34)
8.79 (0.29)
4.99 (0.24)
2.53 (0.08)
14.76 (0.39)
27.51 (0.65)
20.60 (0.60)
2.07 (0.17)
3.48 (0.10)
3.24 (0.11)
13.93 (0.37)
16.03 (0.41)
9.20 (0.29)
11.17 (0.73)
60.15 (2.10)
54.78 (2.99)
5.37 (2.72)
PRIN1
WING
TAIL
W - T
PRIN2
character
0.019
-0.103
0.994
% of total variance
0.626
0.776
0.068
99.2
Skeletal
NASW
MANL
CORL
KEEL
FEL
ULNL
HUML
0.779
-0.621
-0.080
0.6
0.07
-0.830
-0.116
-0.135
0.334
0.274
0.235
0.190
0.078
-0.036
0.214
-0.713
0.411
0.422
0.299
character
0.507
0.174
0.360
0.593
0.313
0.239
0.273
% of total variance
PRIN3
58.5
18.4
8.4
Songcharacter
NPSO
NSSO
DUR
MINS
MAXS
NNA
NFA
MAFA
MAFB
0.079
0.170
0.145
0.210
0.122
0.057
0.788
0.348
0.371
% of total variance
44.0
-0.035
-0.097
-0.017
-0.155
0.072
0.956
0.104
-0.091
-0.161
31.0
0.076
0.318
0.229
0.318
0.166
0.248
-0.595
0.324
0.435
11.7
' Three separatePCAs (external, skeletal, and song analyses)were
performedon specimenand songdata.
All measurementsexceptWEIGHT (g) are in mm.
variance.KEELlengthand NASW hadthe highestloadingson PRIN1. They provided the best
metric studies with PCA, PRIN1 has been inseparationof specimensalong this axis.NASW
terpretedasa "size"axiswhenall the character contributedthe mostto separationof specimens
loadingson PRIN1 are positive(Zink 1988). alongPRIN2. PRIN2 accountedfor 18.4%of the
BecauseTail had a negativeloading on PRIN1, variance. KEEL had the highest loadings on
it is unclear whether PRIN1 is a size axis in this
PRIN3
analysis.
Inclusionof W - T in thePCAisprobably responsible
for this negativeloading.Tail
andWingcontributed
mostto the separation
of
variance.
which
accounted
for 8.4% of the total
SONG CHARACTERS
these taxa on the PRIN2 and PRIN3 axes. PRIN2
Although malesof both taxa sangunivalent
songrepertoires,differencesexistin the syllable repertoireand the pattern of geographic
Skeletal characters.--Overall, the average
Mourning Warbler is larger than the Mac- variation in these syllables.There were five
Gillivray'sWarbler for all skeletalcharacters MacGillivray'ssyllableswhich remotelyresemexceptskull width and tarsometatarsus
length bled Mourning syllables.They were not exact
(Table 2). Resultsof the PCA of skeletalchar- duplicates,differing in syllable morphology,
Geographactersshow that these taxa are essentiallydis- frequency,anddurationparameters.
tinct in multivariatespace(Fig. 4: B). All seven ic variation in Mourning Warbler song is con-
and PRIN3, however, accountedfor only 0.67%
of the variance (Table 3).
skeletal charactershad positive loadings on
PRIN1 which indicates that PRIN1
is a "size"
axis(Table 3). PRIN1 accountedfor 58.5%of the
servativecomparedwith that of MacGillivray's
Warbler.The breeding range of the Mourning
Warbler is dominatedby three regionaldialects.
January1990]
OporornisSystematics
A
167
X: PRINt
Y: PRIN2
Z: PRIN3
[] MourningWarbler
[] MacGillivray'sWarbler
ß Overlap
X: PRINt
B
[] MourningWarbler
[] MacGillivray'sWarbler
ß Overlap
c
Z: PRIN3
Y: PRIN2
X: PRINt
[] MourningWarbler
•
MacGillivray'sWarble
ß Overlap
Z: PRIN3
Y: PRIN2
Fig. 4. Plot of specimens
and songsters
on principalcomponentaxesI, II, III, showingminimal overlap
betweentaxa.Shadedareasrepresentcloudsof pointsoccupiedby individualsin multivariatespace.(A) Plots
basedon analysisof Wing, Tail, and W - T measurements;
(B) plots basedon analysisof seven skeletal
characters;
(C) plotsbasedon analysisof nine songparameters.
In thesedialect systemsall birds sing the same
songtype, differing primarily in the numberof
syllablesand/or physical parameters.In contrast,almosteverymaleMacGillivray'sWarbler
sanga different songtype (Fig. 3: D, E, F). The
total number of syllables encountered versus
the number of birds sampledwas higher for
MacGillivray's Warblers compared with
Mourning Warblers (Fig. 5).
Mourning Warblers sang lower frequency
songson averagecomparedwith MacGillivray's
males (Table 4). Mourning Warblers also sang
lesscomplexsongs.MacGillivray'smalessang
two- or three-part songs,which containedtwo
or moresyllabletypes.In contrast,mostwestern
Mourning Warblers sang one-part monosyllabic songs,whereas most easternmalessang twopart songs.MacGillivray'ssongsaveragedmore
168
I^• PITOCCHI•[U
[Auk,Vol. 107
Columbia at the edge of the boreal forest (Er-
1 oo
skine and Davidson 1976).
Cox (1973) and Salt (1973) observed contact
in central Alberta. Cox (1973) described "con-
acGillivray's
tact points" in central Alberta that involed one
or two individuals from one taxon mixing with
many individuals from the other taxon.On the
arbler
...
Bow River near Kananaskis, he collected a sin-
20
0
o
•o
•o
•o
g
o
ø
•o
•.
•
Number of birds sampled
gle Mourning female mixed in with MacGillivray's Warblers. I collected seven MacGillivray's Warblersin Kananaskis,but I did not
collector observeany Mourning Warblers.Cox
also mist-netted four intermediate (but see Hall
1979)specimens9 km westof RockyMountain
House
and 27 km west
of Caroline
near
Red
Fig. 5. Number of new syllables encountered Deer. ! did not find intermediates or members
with each new songstersampledfor Mourning and of either taxon at the Caroline or Rocky MounMacGillivray'swarblers.
tain House locations. ! also found neither the
pure Mourning Warbler populationsnear Red
Deer nor the pure MacGillivray's Warbler popNFA and NSSO (Table4). Principal components ulations at Trochu reported by Cox (1973).
analysisof songparametersrevealedgoodsepSalt (! 973)found a singleMacGillivray'smale
arationof thesetaxain multivariatespace(Fig. singing among severalMourning Warblersat
4: C). Loadingsof parametervariableson PRIN 1 Pigeon Lake, south of Edmonton,but I found
were all positive(Table 3). The NPSO and NFA only Mourning Warblers at Pigeon and Battle
provided most of the separationof songsters lakes. ! collected both taxa west of Edmonton
along PRIN1. PRIN1 accountedfor 44.0%of the along the AthabascaRiver. ! tape-recordedone
variance.NNA contributed to the separationof MacGillivray'smale and collectedanother near
songstersalong PRIN2, which accounted for
31.0%of the variance.PRIN3 accountedfor only
Hinton. The closestMourning Warblers were
! 1.7% of the variation.
the closest these taxa came to one another
Contributions
from NFA
50 km east of Hinton, north of Obed. This was
dur-
and MAFB were largely responsiblefor separation along this axis.
ing the summerof ! 985.The Mourning Warbler
does not occur in the foothills of the Rocky
Mountains, and MacGillivray's Warblersrarely
HYBRID CONTACT
enter into the foothills
from the mountains.
The
dry lodgepole pine forest of the foothills does
Contact between these taxa is different from
not provide the denseundergrowthfor breedother east-westspeciespairs (e.g. flickers, to- ing required by these taxa. The rare spillover
whees, orioles, buntings), which hybridize in of MacGillivray'sWarblersinto the foothillsand
broad zones acrossthe Great Plains (for review,
beyond,in combinationwith the destructionof
seeRising 1983).Mourning and MacGillivray's boreal forest for farming and ranching in cenwarblers meet irregularly in British Columbia tral Alberta, limits contact between these taxa.
Erskine and Davidson (1976) and British Coand Alberta. In Alberta, Mourning Warblers
breed north of Kananaskis in disturbed second
lumbia Hydro (1981, MS) reportedsimilar conBritish
growth throughout the province (Salt 1973). tact between these taxa in north-central
MacGillivray's Warblers are limited to the Cy- Columbiaat Liard River Hot Springs.I sampled
press Hills in southeasternAlberta and the alongthe AlaskaHighway from Liard River east
with
RockyMountain region from Kananaskissouth to Fort Nelson. ! collected individuals
in southwesternAlberta and along the Red Deer MacGillivray'splumageand songtypesat Liard
River. In BritishColumbia, MacGillivray'sWar- River. The closestMourning Warblerswere ca.
biers are much more common than Mourning 200 km east of Liard River at Steamboat, Kledo
Warblers. They are found throughout British Creek, and Fort Nelson. Erskine (pers. comm.)
Columbia, whereas the Mourning Warbler is pointedout that previousaccountsof Mourning
restricted to the northeastern
corner of British
Warblers at Liard River probably refer to mi-
January1990]
169
OporornisSystematics
T^IlI•E4. Elementarystatisticsof physicalparametersof Oporornis
song.
Mourning Warbler
Character
n
NPSO
NSSO
DUR a
MINS b
MAXS b
NNA
NFA
MAFA b
MAFB b
Duration
variables
188
186
182
187
180
174
133
132
122
MacGillivray'sWarbler
Mean (SD)
n
Mean (SD)
1.72 (0.45)
5.68 (1.05)
1.11 (0.17)
1,933.24 (301.04
5,458.81 (699.44
4.58 (1.36)
2.54 (0.83)
3,975.13 (558.52
3,744.93 (615.41
116
115
112
111
107
100
95
95
85
2.04 (0.62)
6.99 (1.56)
1.29 (0.22)
2,416.05 (334.50)
5,953.18 (648.48)
4.38 (1.42)
4.34 (1.80)
5,437.49 (801.62)
5,388.10 (788.62)
in seconds.
Frequencyvariablesin Hz.
ing mechanism.Vocal charactershave proved
to be useful tools for evolutionary biologists
(Stein 1963,
Springs,thereis a dry lodgepolepine forestthat (Lanyon 1969).Studiesof Empidonax
lackssuitable breeding habitat for either taxon. Johnson1980)and Myiarchusflycatchers(LanThis acts as a barrier to contact between these
yon 1978)have shownthe value of songchartaxa.
actersin delimiting sibling taxa. Major differFurther contact between these taxa seems unences in primary song also occur between
likely. In BritishColumbia,the breedingranges Mourning and MacGillivray's warblers. Aldo not come closetogether. Agricultural prac- though there is geographicvariation in songin
ticesin Alberta have been largely responsible both taxa, there is no syllable sharing between
for widening the gap between Mourning and these taxa, and the pattern of geographic variMacGillivray'swarblers.Beneaththe foothills ation in thesesyllablesis different in eachtaxof the RockyMountains,farming and ranching on. Almost every MacGillivray's male sang a
are destroying suitable breeding habitat for unique song.In contrast,the breeding range of
Mourning Warblers and thus pushing the the Mourning Warblerwasdominatedby three
breedingrangeof the Mourning Warbler east- major dialect systems(Pitocchelli 1988). Furward and northward away from any potential thermore, there is evidence of song displacecontact.
ment in thesetaxa.The westerndialectsystem
of Mourning Warbler males is dominated by
DISCUSSION
one-part monosyllabic songs, whereas Macgrants or vagrantsbut not to breeders.For 200
km between
Fort Nelson
and Liard
River
Hot
Species
limits.--These taxa overlap in plumage
characters,but the incidence of overlap is low.
Gillivray's male sing two- and three-part songs.
Two-part songsdominate the eastern dialect
Very few MacGillivray's Warblers resemble systemsof the Mourning Warbler. Basedon
Mourning Warblers, and then only for the principal componentanalysesof physicalpawing-minus-tail(W- T) character.Although rametersof song, operational taxonomicunits
manymoreMourningWarblerspossess
eye-arcs, (OTUs) of each taxon occupya unique portion
dark lores, or (W - T) < 11 mm, none of the of "songspace."Mourning maleswith two-part
intermediate Mourning specimens ! studied songsuse different syllablesand occupysepapossessed
all threeMacGillivray'scharacters.
In rate songspacefrom MacGillivray's maleswith
contrastto plumage, PCA of the external and two-part songs(Fig. 4: C). Song differencesare
skeletalmeasurementsshow separationof these strongand consistentthroughoutthe rangesof
taxa (Fig. 4: A, B). Becausethese taxa do not these taxa. Song types of eastern and western
hybridize in large zones, the intermediacyof Mourning Warbler specimensthat were interproblem specimenscannot be due to hybrid- mediatefor plumage characterswere similar to
ization events.Intermediateindividuals simply "normal" plumage neighbors.Mourning Warfall within the normal range of variability for bler maleswith eye-arcsor dark loressangthe
each taxon.
samesongsas Mourning Warblersthat lacked
Bush (1975) emphasizedthe importanceof thesecharacters(Fig. 3: A, B, C). MacGillivray's
bird songasan exampleof a prezygoticisolat- Warblers which resembledMourning Warblers
170
1^¾PITOCCHELLI
for the W - T measurementdid not have songs
with Mourning Warbler syllables.This problem
is similar in other sibling specieswhere behavioral differentiation may have proceededfaster
than morphologicalcharacters.Although these
communicationssystemsappearunique to each
taxon,the role of learning versusthe influence
of behavior on songhas not been investigated.
The central questionof my study was whether to ascribespecificstatusto these taxa or combine them into a single taxonastwo subspecies.
Traditionally, interbreeding has been the most
important criterion for making this iudgment
(Mayr 1969),but someauthorsstressgenotypic
and phenotypic distinctnessover the ability to
[Auk,Vol.107
below the eye. Eye-arcsin Mourning Warblers
are usually weak, but someextremespecimens
may resemble full eye-arcsof MacGillivray's
Warblers.If, in spring, males eastof the Rocky
Mountainsdo not possess
eye-arcsanddark lores,
then they are Mourning Warblers. If they
possessboth, then refer to the song type (if
possible) and the W- T measurement. AlthoughI collectedfemales,the samplesizeswere
much lower than in Lanyon and Bull (1967).
Therefore, for identification of females, refer to
the quality of the eye-arc (thick versusweak)
and Lanyonand Bull'sW - T measurementfor
separation.
interbreed (Cracraft 1983, McKitrick and Zink
1988). In my opinion, Mourning and Mac-
ACKNOWLEDGMENTS
I am extremely grateful to my doctoralcommittee:
Gillivray's warblersshould continue to be consideredseparatespecies,basedon the distinct- W. E. Lanyon,R. Rockwell,F. Gill, P. Mundinger, and
M. Hecht. G. Barrowclough,L. L. Short,and L. Marcus
ness of their primary song and skeletal
alsohelped formulatethis project.I thank R. W. Dickdifferences.Although their plumagesare sim- erman, who helped with specimenpreparation.I apilar, operational taxonomicunits of these taxa preciatethe use of the computerfacilities at Queens
occupydifferentportionsof morphologicaland CollegeLaw Schooland City Universityof New York
songspace.The different patternsof geographic Graduate Center. Peter Capainolo assistedme with
variation in primary song--conservative in
Mourning Warblersversushighly variable in
MacGillivray's Warblers--also point to separate
evolutionary histories.These taxa are also still
essentiallyallopatric,and hard evidenceof assortativematingis unattainable.Resultsof preliminary playbackexperimentssimulatingsymparrybetweenthesetaxarevealedthat malesof
both taxa can discriminateconspecificfrom experimental song types (Salt 1973, Pitocchelli
1988). Only 1 of the 25 Mourning and MacGillivray'swarblerswasunable to discriminate
between song types. This supportsthe importanceof the distinctnessof primary songin these
taxa and alsopoints to positive assortativemating if and when these taxa come back together.
the fieldwork in New York. The manuscriptwas improved by the insightful commentsof G. Barrow-
clough,R. W. Dickerman,and F. Heppner.Funding
for this projectwas provided by the F. M. Chapman
Memorial Fund (American Museum of Natural His-
tory), Undergraduate-Graduate Research Award
(American Museum of Natural History), American
Museum Predoctoral Fellowship Award (American
Museum of Natural History), Alexander Bergstrom
Award (Northeastern Bird-Banding Association),
Alexander Wetmore Award (American Ornitholo-
gists'Union), SigmaXi ResearchAward, (Sigma Xi
Society),and institutional funding from the City Uni-
versityof New York. I alsoappreciatethe generosity
of F. Haas, B. Pitocchelli,E. Mangan, and the Mislenovichfamily. Finally I thank the CanadianWildlife
Service(regional officesin British Columbia and Alberta), U.S. Fish and Wildlife Service, and the follow-
Identification
of extreme$pecimen$.--Identifi- ing state Game and Fish agenciesfor taking time to
cation of extreme specimenscontinues to be a
problem for this complex.Exceptfor songtype,
no singlecharacterdistinguishesthesetaxa100%
of the time, and songis uselessfor femalesand
migrants. Although W - T appears to be the
best morphological character, my results and
Kowalski's (1983) have shown that it is equivocal for separating some extreme specimens.
Most problemspecimenshave been Mourning
Warblersthat resembleMacGillivray's Warblers
approveand issuethe appropriatecollectingpermits:
Washington, Oregon, California, Idaho, Nevada,
Montana, Wyoming, Ontario, Quebec,New York, and
Maine.
LITERATURE CITED
AMERICA•q ORt,IITHOLOGISTS' UNIOiXl. 1983.
Check-list
of North American birds. Sixth ed. Washington,
D.C., Am. Ornithol. Union.
BAIRD,$. F. 1858. PacificRailroadSurveys,vol. 9,
for one but usuallynot all plumagecharacters.
part 2. Washington,Beverly Tucker Printers.
Combinations
of these characters are useful for
BAPTISTA,
L. F. 1977. Geographicvariationin song
identifying extreme specimens.Full eye-arcs
dialectsof thePugetSoundWhite-crownedSpar-
containthick layersof white feathersaboveand
row. Condor
79: 356-370.
January
1990]
Oporornis
Systematics
171
BECKER,
P. H. 1982. The coding of species-specific MENGEL,R.M. 1964. The probable history of species
formation in some northern wood warblers (Pacharacteristicsin bird sounds. Pp. 214-252 in
Acoustic communication in birds, vol. 1 (D.
rulidae). Living Bird 3: 9-43.
Kroodsma and E. H. Miller, Eds.). New YorkßAcPAtti, S. T., & M. L. MEYERS. 1976. A probable
ademic Press.
Mourning x MacGillivray'swarbler hybrid. Wilson Bull. 88: 490-491.
BEIMBORN,D. 1977. Possible MacGillivray's x
A. R. 1947. The races of MacGillivray's
Mourning warbler hybrid. PassengerPigeon 39: PHtILLI?S,
Warbler.
257-258.
BUSH,G.L. 1975. Modesof animal speciation.Annu.
Rev. Ecol. Syst. 6: 339-364.
CHAPMAN, F.M.
1917. The warblers of North Amer-
ica. New York, Dover Reprint Ed. (1968).
Cox, G.W. 1973. Hybridization between Mourning
and MacGillivray's warblers.Auk 90: 190-191.
COUES,
E. 1903. Key to North American Birds.BosCRACRAFT,
J. 1983. Speciesconceptsand speciation
analysis.Pp. 159-187in Current ornithology,vol.
1. (R. F. Johnston,Ed.). New York, Plenum Press.
ERSKINE,A. J., & G. S. DAVIDSON. 1976. Birds in the
lowlands
of northeastern
British Co-
lumbia. Syesis9: 1-11.
HALLßG. 1979. Hybridization between Mourning
and MacGillivray's warblers. Bird-Banding.50:
101-107.
HOFSLUND,
P. B. 1962. The genusOporornis.
Flicker
34: 43-47.
JOHNSON,N.K.
York.
RISINGß
J. D. 1983. The Great Plains hybrid zones.
Pp. 131-157 in Current ornithologyßvol. 1. (R. F.
JohnstonßEd.). New York, Plenum Press.
ß & F. W. SCHUELER. 1980. Identification
ton, Dana Estes and Co.
Fort Nelson
Auk 64: 296-300.
PITOCCHELLI,
J. 1988. Charactervariation in the Oporornisphiladelphia-Oporornis
tolmieicomplex.Ph.D.
dissertation.New YorkßCity University of New
1980. Character variation and evo-
and
statusof Wood Pewees(Contopus)
from the Great
Plains:what are sibling species?Condor 82: 301308.
ROBBINSß
J.D., & G. SCHNELL.1971. Skeletalanalysis
of the Ammodramus-Ammospiza
grasslandsparrow
complex:a numerical taxonomicstudy. Auk 88:
567-590.
ROHWER, S. A.
1972.
A multivariate
assessment
of interbreedingbetweenthe meadowlarksß
Sturnella.Syst. Zool. 21: 313-338.
SALTßW.R. 1973. Alberta vireos and wood warblers.
Publ. 3, Prov. Mus. Arch. Alberta.
lution of siblingspeciesin the Empidonax
difficilis- SAS Institute. 1985. SAS user'sguide: statistics(version 5.16). Cary, SAS Institute Inc.
fiavescens
complex.Univ. Calif. Publ. Zool. vol.
112.
SHIOVITZ,K.A. 1975. The processof species-specific
KOWALSKI,
M.P. 1983. Identifying Mourning and
songrecognitionby the Indigo Bunting, Passerinacyanea,and its relationshipto the organization
MacGillivray'swarblers:geographicvariation in
of avian acoustical behaviour. Behaviour 55:128the MacGillivray's Warbler as a sourceof error.
North
American
Bird
Bander
179.
8: 56-57.
LANYON,W. E. 1969. Vocal charactersin avian systematics.Pp. 291-310 in Bird vocalizations(R. A.
Hinde, Ed.). Cambridge,CambridgeUniv. Press.
1978. Revision of the Myiarchusflycatchers
of South America.
Bull. Am. Mus. Nat. Hist.
161:
.,& J. BULL. 1967. Identification of Connecticut,
Mourning and MacGillivray's warblers. BirdBanding 38: 187-194.
MACSPIN. 1988. Graphical data analysisand decision support software (version 2.0). Austin, D 2
Softwareß Inc.
25.' 223-226.
York, McGraw-Hill
TAVERNER,P. A. 1919. The birds of the Red Deer
Riverß Alberta. Auk 36: 1-21, 248-265.
THORPEß
R.S. 1976. Bitmetric analysesof geographic
variation
and racial affinities.
Biol. Rev. 51: 407-
452.
pollsCarduelis
fiammea
fiammea
andC. hornemanni
exilipes.Auk 102:82-96.
MAYR,E. 1969. Principlesof systematic
zoology.New
Book Co.
birds. Publ. Nuttall.
ZINK, R. M.
1988. Evolution
of Brown Towhees: al-
lozymes,morphometricsand specieslimits. Con-
& L. L. SHORT. 1970. Speciestaxa of North
American
Co.
TROY,D. M. 1985. A phenetic analysisof the Red-
MARCUS,
L. 1982. Aportable, digitalprintingcaliper.
--,
man
STEINßR. C. 1963. Isolating mechanismsbetween
populationsof Trail's Flycatchers.Proc.Am. Phil.
Soc. 107: 21-50.
429-621.
Curator
SOKAL,R. R., & P. H. A. SNEATHt.1963. Principles of
numerical taxonomy. San Francisco,W. H. Free-
Ornithol.
Club. 9.
McKIVRICK,M., & R. M. ZINK. 1988. Speciesconcepts
in ornithology. Condor 90: 1-14.
dor 90: 72-82.