The Auk 113(1):105-118, 1996
NEOTROPICAL
BIRD MIGRATION ORIENTATION AND
DURING THE ICE AGES: ECOLOGY
TIMOTHY C. WILLIAMS t AND THOMPSON WEBB III 2
aDepartment of Biology, Swarthmore College Swarthmore, Pennsylvania 19081,USA;and 2Department of Geological Sciences, BrownUniversity, Providence, RhodeIsland02912,USA
ABSTRACT.--Reconstruction of breeding habitat of North American Neotropical migrants 18,000yearsagoand 9,000yearsago indicatedmajorshiftsin both locationand composition of plant communitiesrelativeto presentconditions.Increasedvegetationin xericareasmay have compensated,at least in part, for the reduction in breeding habitat due to glaciation. Autumnalflightsof Neotropicalpasserinemigrantsflying on constantheadingsfrom North America to Central and South America were simulatedunder presentwind conditionsand for windsduring periodsof glaciationat 18,000and 9,000yearsago.The 155øaverageheadings currently observedfor Atlantic migrantswere found to function well during periods of glaciationand may have been more generallyusefulduring thosetimesthan at present. Received 15 June1994,accepted 27 January1995.
CYCLES OFQUATERNARY GLACIATIOIq during the last 1.6 million years have influenced the development of present routes of both Neotropical and Paleotropicalmigration (Moreau 1972, Gauthreaux1982).North AmericanNeotropical migrantswere mostclearlyaffectedbecausemost of their presentbreeding groundswere covered by a permanent ice cap only 18,000 (radiocar-
bon)yearsago(18 kya) and becausethe ecology of North America hasundergoneseveralmajor changesduring the period of deglaciation(Imbrie and Imbrie 1979, Emery et al. 1988, Webb 1988).Modern speciesassemblages or plant formations are of recent origin in North America. Most developedafter early Holocenewarming:
modern tundra first appearedat about 8 kya, borealforestat 7 kya, and modernmixed forest at 6-8 kya. Only modernprairie, first appearing about 10 kya, and modern deciduousforest (12 kya) existedduring periodsof significantglaciation (Webb 1988). After their first appear-
ing the fitnessof current bird populationsover their shortlifetime?Would the orientation system(s)used by birds under present conditions have been effective under past conditions? We examine whether fixed-headingorientation would be sufficient to guide Neotropical migrants during periods of glaciation. We first estimate breeding habitats of Neotropical migrantsin North Americaat 18 kya (full glaciation) and at 9 kya (partial glaciation).We then simulatemigratory flights at fixed headingsfor these birds in order to examine
the constraints
posedon orientation at those times in the past. The Neotropical migrants we discussare a sub-
setof the type A Neotropicalmigrantsdefined by the ResearchWorking Group of Partnersin Flight (1992). We consider only autumnal (southward)migration. We excludediurnal migrantssuchas raptors,which are clearly influenced by local topography,and excludebirds
ance, plant formations increased in area and
moving southwestwithin North America. We focuson shorebirdsand passerinesflying south
approachedtheir presentdistributionsonly in the past4,000years(Van Devender 1986,Webb
South America.
1988). The present distributions of plant taxa were relatively rare events during the Quaternary and, like the interglacial periods with which they areassociated, haveexistedfor only 10%of the 100,000-yearglacial/interglacialcycles (Ruddiman and Raymo 1988). These findings posetwo majorproblemsfor the study of the evolution of bird migration. What behavior would allow bird speciesto track changes in habitat over thousandsof years,while preserv105
to southeast
from North
America
to Central
and
We refer to this subset of birds
as Neotropical migrants. Severallinesof evidence,including radarand radio-telemetryobservationsof migration, the experimentalmanipulationof captivebirds,and
simulatedmigratoryflights,suggestthatat least somesongbirdsand shorebirdsguide long-distance migrations by maintaining a fixed compassheading. Drift by winds producesa variable track, but by selecting synoptic weather conditionsfor take-off,thegreatmajorityof birds
106
W•LL•^•S^•D WœBB
can move reliably between breeding and wintering areaseachyear (Williams and Williams
[Auk, Vol. 113 METHODS
Cochran and Kjos 1985, Alerstam et al. 1986,
We usedthe mappedpollen datafrom Webb (1988) and Overpeck et aL (1992) to estimatebird habitats. We labeled these habitatsas desert, prairie, tundra,
Richardson 1991, Williams 1991). Dire ction and
coniferous forest, mixed forest, and deciduous forest
somecue for the cessationof migration (dis-
to describestructuralfeatures,but do not imply present speciesassemblages for each region in the past (Baker1983,Webbet al. 1987,Thompson1988,Webb 1988). We set the southernlimit of bleeding habitat
1978, 1990, Alerstam 1981, Stoddard et al. 1983,
tance, time, latitude, or habitat) appear to be
geneticallyencoded(Wiltschkoand Wiltschko 1978, Gwinner 1986, Berthold 1988, 1990). One
or more such vectors direct first-year birds to the general location of stopoverand wintering areas,where they then searchfor suitablehabitat. In subsequentyears, birds return to these local habitatsusingfamiliar landmarksor some other orientation systemoperationalwithin the areasspecifiedby the codedvector (Wiltschko and Wiltschko 1978, Williams and Williams 1990,
R. Wiltschko1992).A varietyof approaches have shown that suchfixed headingscouldbe maintained by use of a compassbasedon the sun, stars,wind, or magneticfield of the earth(Able 1980, Alerstam 1981, Wiltschko and Wiltschko, 1988).
Although birdsmay useother techniquesfor
for Neotfopicalmigrantsasthe southernlimit of killing frost and set no northern continental limit. Bleeding habitatincludesall iofest types,wetlands,prairie and tundra, but excludesdesert and permanent ice. We consideredArctic coastaland North Slope areas assuitablehabitatfor shorebirds,but not Neotfopical passefine migrants. To simulate migratory flights, we first created a wind matrixfor North Americarepresentative of those winds actuallyusedby Neotfopicalmigrants.We then
simulated•lights through thesewind fieldswith an iterative computer program.Finally, we modified the wind matrix to estimateconditionsduring glaciation and repeated the simulations with these wind conditions.
The methods
for creation
of a wind
matrix
under
1991,
presentconditionsare describedin detail in Williams (1991).The 5ølatitude x 5ølongitude matrix for present conditionswasan averagewind velocityfor those
Wallraft 1991, R. Wiltschko 1992), orientation
North American weather systemsknown to have sup-
orientation (Rabol 1978, Able 1980, Wiltschko and Wiltschko
1988, 1991, Richardson
by a fixed compassheading is consistentwith availabledatafor Neotropicalmigrantsmoving south
to southeast
across North
American
to
ported maiof autumnal migrationsof passefinesand shorebirdsmoving to the south and southeastas observedby radar.Weather systemsassociated with such movements
were determined
from radar observations
Central and South America (Williams et al. 1977a, 1977b, Williams and Williams 1978, 1990,
in: Alberta (Richardson and Gunn 1971); Wisconsin
Williams 1985).Radarobservationsof Neotropical migrantsoverthe Caribbean,Bermuda,and
to Office of Naval Research); Ontario and Quebec
over the North
American
eastern coast revealed
averageheadingscloseto 155ø(relative to true north) at all siteswith no significantchangeof headingsto compensatefor wind drift (Williams and Williams 1978, Williams 1985). In simulations
birds move
southeast
across North
America and over the western North Atlantic,
where they encounternortheasttrade winds that drift them westward
to Central
and South
America.The bestavailablesampleof the radar observationsis from Antigua in the Caribbean. At Antiguathe averageheadingwas 153ø, with an angular (standard)deviation of 26.9ø (Williams and Williams 1978, Williams 1985). Mean
(T. C. Williams and J.M. Williams unpubl. 1979report (Richardson 1972); and along the North American easterncoast(Dfury and Keith 1962,Richardson1972, 1979,1980, Williams et al. 1977a, 1977b, Williams 1991).
At all sites, south of southeast migrations likely to
containsignificantnumbersof Neotropicalmigrants occurredwest of a cold front, when barometricpressurewas rising of steadyand when winds were northerly, westerlyof calm(seeRichardson1978).The concurfence
of these factors was then considered
to be
criteriafor identifying weather systemslikely to produceNeotfopical migratory flight in areasof at times when we lacked radar observations.
South and south-
eastmigration that occursunder other, less-predictable conditions was excluded (Richardson 1978, 1990). The wind
matrix for current
conditions
was created
by averaging 774 wind measurementstaken at 850 mb (ca. 1,500 m altitude), and 595 surface-wind measurements.This gave an estimateof winds close to the averagealtitudeof migrantbirdsovercontinental
headingsobservedat other windward islands, Bermuda,and Miami ranged from 151ø to 157ø (Williams 1985).Airspeedsat Antigua showed areas(Eastwood1967,Able 1970,Williams et al. 1977a). two peaks,one at 45 kin/h, presumably passer- Wind data from 500 mb (ca. 5,000 m) were used to ines, and one at 75 kin/h, presumably shore- modifysurfacedatawhen 850 mb datawere lacking. birds (Williams 1985).
Flight at 5,000m is fare over continentalareas,al-
January1996]
IceAgeBirdMigration andEcology
though it has been documentedfor shorebirdsinitiating transoceanic flight (Richardson1979).Flight at thesealtitudes at a constantheading has been simulated over easternNorth Americaby Williams (1991), but is not included
in our simulations.
Matrix valuesfor presentwinds in areasnot penetratedby North Americanweathersystemswere de-
107
etation and faunas without analog in today's plant and animal communities(COHMAP 1988, Overpeck et al. 1992). Analyses of past pollen distributionsshow that the major tree taxa respondedindependently to past changesin cli-
mate.Majorpollen-producingtreesin both Eu-
Atlantic were averagewind velocities,recordedat the altitude of birds, during radar observationsof moderateor heavyautumnalbird migration in thoseareas
rope and North America show changesin distribution that are independent of other taxa (Huntley and Webb 1989). The scant data on Quaternaryavian distributionalsoindicatethat
(Williams et al. 1977a, 1977b, McClintock et al. 1978,
birds utilized
Williams 1985). Prevailing winds were used over
rent speciesassociations (Gauthreaux1980).
termined
as follows.
Winds
over the western
North
oceans south of 30øN. We limited our wind matrix to
latitudes 10øSto 70øNand longitude 50øWto 130øW becausewe lacked adequate radar observationsin South America
and the Arctic.
Parametersfor heading and airspeedused in the
habitats dissimilar
from their cur-
Since associations among plant taxa were
ephemeral during the Quaternary, we used physicalandstructuralfactorsto identify breed-
ing areasfor long-distance migrants.Summers
had to be sufficientlywarm and wet to provide a large insectand fruit food source,and winters sufficientlycold to greatly reduce these food sources.Severeweather conditionsduring the m/s), about one standard deviation below the mean nonbreeding seasonin these areas would seairspeedof late season(passerine)migrantsobserved verely limit the residentpopulationsof birds, with radarat Antigua,and representsthe slowestmi- mammals and parasitic insects, thus reducing grantslikely to make the Atlantic crossing(Stoddard competition, predation, and parasitism for et al. 1983). breeding migrants (see Cox 1985). Below we Winds at 9 kya and 18 kya were estimatedby modgive habitat and migratory-route reconstrucifying the presentwind-velocity matrix in relation to the National Center for AtmosphericResearchCom- tions for 18 kya and 9 kya and comparethese with the present conditions. In each casewe munity Climate Model (NCAR CCM) simulationfor begin with a descriptionof simulatedclimate North Americaat 9 kya and 18 kya as given in Kutzbach (1987). NCAR CCM simulations were available and paleoecology to locatelikely breedinghabsimulations were derived from more than 2,000 radar
tracksof migrantsobtainedover Antiguafor five migratoryseasons. Forthe simulationwe useda heading of 155ø + 25ø. Simulated airspeedwas 35 km/h (10
for Januaryand July for theseperiods.A model for itat and then consider the effects of constant modificationof the simulatedwind patternswasob- heading orientation on Neotropical migration tained by comparisonof presentwinds observeddurunder these conditions. ing southeast migrationswith presentsimulatedstorm 18,000years ago.--Glaciationhas been the tracks,high- and low-pressure centers,and resultant predominant climate condition for North winds for Januaryand July.Simulatedresultantwinds
alongthe presentsimulatedJanuarystormtrackapproximated winds observedduring autumnal migra-
tion more closelythan did thoseof presentJulysimulations.The averagelocation of Januaryand July high-pressure centersand the averagevelocityof January and July trade winds correspondedto thoseobserved during presentautumnal Neotropical migrations. We obtainedJanuarystorm tracks,averagelocationof stabilehigh- and low-pressurecenters,and resultant winds for January and July, from NCAR CCM simulations for 9 kyaand 18kyaandsubstituted thesevaluesin the model derived from presentsimulations to obtain a 5ø x 5ø matrix of estimated winds
usedby autumnalNeotropicalmigrantsat 9 kya and 18 kya. RESULTS
The climatesof glaciatedNorth Americawere unlike any todayand this conditionled to veg-
America for more than 80% of the past900,000 years(Ruddimanand Raymo1988).At the height of glaciationmuchof the presentbreedingarea of Neotropicalmigrantsnorth of 45øNwascovered by ice up to 3,300 m thick (Fig. 1C). The principal exceptionswere Alaska which was largely ice-free, and possibleice-free areas in the Canadian far north and between the eastern and western ice domes. Both summer and win-
ter temperaturessouthof the ice were 2øto 10øC lower than present.Unglaciatedareasin Alaska, as well as in northwestern, north-central, and southeastern United States, were drier than
at present;the Northeastand Southwesthad wetter summersthan present (Kutzbach 1987, COHMAP 1988).
Lowering of sealevel by about ! 00 m exposed large areasof new land. We suggestthat substantial
additional
habitat
was available
to Neo-
108
WILLIAMS ANDWEBB
A
•
B
500 ya
-• Arid areas
• "•=•'•'•:;• 9 kya
[Auk,Vol. 113
C 18 kya
__ Ice ,• Shorebird habitat only Breeding habitat
Fig. 1. Estimatedbreeding habitat for passerineand shorebird Neotropical migrants based on criteria given in text. Stippledareawould supportmajorshorebirdbreedingpopulations,but probablynot passetines. Northern extent of breeding areas9 kya and 18 kya assumesice-freeArctic coastalareasand tundra structure similar to present.At 18 kya, land area was continuouswith easternSiberiaand ice cap extendednortheast to cover Greenland; these areashave been truncated.Margins of ice capsand coastlinesfrom Denton and Hughes (1981)and from COHMAP (1988).Glaciallakesshown:Agassiz(9 kya), and Lahontanand Bonneville (18 kya).
tropicalmigrantsalong the eastand west coasts land (Jacobsonet al. 1987). Northwestern North of North America, and in the Caribbean and America consistedof a subalpine open woodGulf of Mexico (seeFig. 2C). The extentof this landgradinginto tundrain ice-freeAlaska(Barhabitat is difficult to estimate as areas near the nosky et al. 1987, Engstrom et al. 1990). In ice would have been depresseddue to its mass mountainousareas,the tree line was depressed (Denton and Hughes 1981, Peltier 1987), and 1,200 to 500 m, with alpine vegetation below Caribbean areaswere subjectto tectonic forces glaciatedsummits(Baker 1983).The increased that may have depressedland areas(Aubry et precipitation and lower temperatures in the al. 1988). As suggestedby Emery (1967), the Southwestcreatedlarge shallow lakes,suchas nature of these lands either could have been Lake Bonneville (Fig. 1C). Present scrub and estuarine, where sea level dropped below the desert areas in the United States were a mixed continental shelf, or bay and wetlands, where woodland.The Mexican highlandswere drier the seaswere shallow.Vegetationin theseareas than at present(Baker1983,Van Devender 1986, of the exposedcontinental shelf was similar to COHMAP 1988). The total breeding habitat available for Neothat on nearby continental margins (Emery 1967). tropicalmigrantsappearsto havebeenlessthan Pollen distributions from eastern North at 500yearsago (ya) due to glaciationof almost America 18 kya show a great reduction in de- all of the presentnorthernbreedingrange(Fig. ciduousforestscomparedto present.A pine for1). This lossmay have been compensatedsomeest with some deciduous trees covered the what by the addition of excellent habitat in southeasternUnited Statesand probably much greatly expanded coastalareas, in areas near of the area of the exposedeasterncontinental pluvial lakes, and in the increasedwoodland shelf (Emery et al. 1967,Jacobsonet al. 1987). areas of the Southwest. With reduced opporAn open woodlandwith deciduoustrees,grass- tunities for migration, some speciesmay have esand prairie forbscoveredthe areaof the Unit- become resident in expanded Caribbean land ed Statesgulf states.Sedgesandforbsindicating areas.During periodsof maximumglaciation, open to treeless vegetation, dominated the speciesthat at present breed in boreal forest northeastern United States. South of the ice in may have been able to competesuccessfully in the midwesternUnited States,enough spruce the generally coniferous,cooler habitatsover treesgrew to yield a vegetationsimilar to park- much of the continental United States. There
January 1996] 130
120
110
IceAge Bird Migration and Ecology 100
90
80
70
60
50
130
120
110
100
109 90
80
70
60
50
i
70-1
(50-! 50--• 40--
•
•
30--
20--
_,•
•
•0
ß•,•
155
•
•
-•0
'•,•.
A Present
10-130
120
•10
100
90
80
70
60
i
i
I
I
i
i
I
S0
130
120
Ice•
1•0
•00
90
Breedin9 •
habitat
80
70
60
SO
Wind ', ', 20
-60
130
120
110
100
90
80
70
60
50
Fig.2. Simulated migration routes andbreeding habitats forhypothetical Neotropical migrantpasserine movingfrombreeding grounds in NorthAmerica to winteringareasin northernSouthAmerica at three timeperiods: Present, 9 kya,and18kya.Simulated tracks of birdsflyingat airspeed of 35km/h(10m/s) withfixedheadings asindicated; position plottedat 1-hintervals. Representative windsof the5ø x 5øwind matrixusedforsimulations shownby solidarrows.(A) Forpresentconditions, windsareaverage forsynoptic
conditions selected bypresent autumnal migrants movingsoutheast overeastern NorthAmerica. Breeding habitatshownmeetscriteriadescribedin text.Forsimulations at (B)9 kyaand(C) 18kya:windsandbreeding
habitats estimated fromglobalclimate simulations; areaoficecapfromDentonandHughes (1981). All panels showtracksof birdspassing overislandof Antiguaon average headingobserved in Caribbean (155ø)and for 130ø and 180ø. Tracksalsoshown for birds using headingsof 155ø from extremeeasternand western
portions ofbreeding area.Additional breeding areas exist(ed) northandwestofgeographical limitsoffigures. Degrees of latitudeandlongitude givenat panelmargins. Mercator projection distorts relativeareasand distances,but showsdirection of tracksand winds accurately.
110
WILLIAMS ANDW•ss
[Auk,Vol.113
may have been considerablehabitat for birds suchas BlackpollWarblers(Dendroica striata)in Alaska 18 kya, depending on the northward
by the ice cap. A large high-pressurecenter dominatedwinds over centralCanadacreating relatively light wind in this area.The average
extension of conifers into the tundra of that
storm track split into two. A northern track, archedup over north-central Canadaand down
periodand the compositionof the tundra,which
differedfrom that of today(Webb 1988,Engs- along the northeasterncoast,creating strong tromet al. 1990).Much of the Nearcticwasprob- northwest winds in northeastern Canada. A ably ice free in summerand wassuitablebreed- southern branch of the storm track crossed the ing habitatfor nonpasserine migrants(Johnson southern United States at about 40øN and then and Herter 1990). turnednorth alongthe coastto join the northTo facilitatecomparisons of pastand present ern track. The southern storm track created migration routes, we simulated the breeding
strongnorthwestwinds behind coldfrontsover
areaof a hypotheticalNeotropicalmigrantpasserinethat todaymovesfrom Canadaand Alaska to Southand Central America(Fig. 2). The breedingarea of this migrant is similar to that of the taxa that Mengel (1970) termed taiga wood-warblers. Alaskanareas(shownin Fig. 1) are not included in Figure 2. At 18 kya (Fig.
most of North America between 30øN and 50øN
2C), the breeding area was shifted far to the southof the presentdistribution(Fig. 2A). The
drifted
(Fig. 2C). Thesewinds were more favorablefor southeastmigration in the western and central United Statesthan are present wind conditions
(Fig.2A; Kutzbach1987).Birdscontinuingfixed heading flight over the Atlantic encountered northeast trade winds south of Bermuda, which them
westward
toward
Central
and
northeasternportion of the range of the hy-
South America. Simulated migrants moving along the west side of Central America found
pothetical migrant was restricted to a narrow
similar
strip of exposed continental shelf between the
drifted
ice and the sea.An alternativeinterpretationof breeding areais to restrictthe wood-warblerto
the muchsmallerareasof sprucelocatedin the
(Fig. 2C). At 18 kya both passetinesand shorebirds passingover Antigua could have used the same
midwestern
constantheadingsas they do today (Fig. 2 A
states.
In this and the following simulations,we first
investigatewhether the headingsof 155ø + 25ø
but less extensive west winds which them eastward toward South America
and C). Passerineson headingsof 155ø to 180ø would
have
arrived
from
the narrow
north-
thatreliablybringbirdsoverAntiguaontheir easternseacoast,asthere were no suitablepasway to South America at present would have been functional in the past.We alsoplot tracks with a headingof 155ødepartingfrom the western and eastern limits of the hypothetical breeding range to see if these headings are broadlyapplicablein North America.In Figure 2 we usean airspeedof 35 km/h (10 m/s), equal to the slowestpasserinescurrently migrating over Antigua. Previous work has shown birds with low airspeedsto be the most constrained in their use of fixed-headingorientation (Stoddard et al. 1983, Williams 1991). We also simulated flights at airspeedsof 75 km/h (21 m/s) near the averagefor shorebirdsflying over Antigua.Thesesimulatedtracks(not shownin Fig. 2) were lessdeflected from their heading and took lesstime to completetheir flight (seeStod-
sefine habitats north of 50øN in eastern North
America.Passerines on headingsof 130øcould havecomefrom asfar asAlaska.Birdsbreeding north of the ice 18kya would maketwo nonstop migrations. The first, over the ice cap, would take about 90 h for the slowestpasserines(35 kin/h) and about 50 h for shorebirds (75 kin/
h). The over-waterflight wouldhavetakenabout the sametime asat present,up to 110h for slow passerinesand 60 h for shorebirds.
Of particularinterestare flights from Alaska on headings centered around 155ø (the most
westerlytrack in Fig. 2C), and flights from the areabracketedby the easternand western155ø tracksin Figure 2C. Theseareasconstitutedthe
greatmajorityof breedingarea18 kya, and Figure 2C showsthat fixedheadingflightsat about
dard et al. 1983,Williams 1991).Tracksare shown
155ø,coupledwith favorablewinds at 18 kya,
from the most northerly point, but tracksinitiated from a more southerly point will follow the samepath, given that there is no effect of previousexperiencein the simulation. Wind patternsat 18 kya were greatly affected
could have provided a rapid and efficientroute to wintering areas in South America, Central Americaand the Caribbean.For the great majority of migrantsto SouthAmerica,thesetracks
would have taken birds over the (probably)
January1996]
IceAgeBirdMigrationandEcology
111
much expandedland areasof the Caribbeanand Central America,a logical initial route between North and South America. Migrants at this time might have used fixed headings not to make long flights over open ocean or inhospitable
packrat(Neotoma)middens from western North Americaindicate closed-canopyconiferousforestsin coastalregionsand at higher elevations of the northwest, with grasslandsin the drier lower elevations (Baker 1983). The flora of the
arid lands, but to direct movements over well-
southwestern
vegetated continental areas or to guide relatively shortflightsover water betweenland ar-
present,but forestsextendedfurther upward in altitude due to warmer temperatures and fur-
eas.
ther downward
9,000 yearsago.--Nine thousand years aõo a reducedNorth American ice capcoverednortheasternNorth America(Fiõ. 115),but wasrapidly recedinõ.As perihelion (closestapproachto the sun) occurredin July, summerswere warmer than at present except for areas near the ice. Simulated Januarytemperaturesindicated the distributionof killinõ frost would be similar to the present(Kutzbach1987).Ice-free coastalareas probably existed alonõ the northern and easternmarõinsof the ice due to the advection
habitat covereda larger area than at presentin most areas (Thompson 1988). Deserts, such as the ChihuahuaDesert,were grasslandsdue to
es were found in central North America, but
less than 45 h to cross the ice, even for the
were associatedwith pines and sprucesin the
slowestpassetines(35 kin/h), while the overwater flight would takeup to 110h asat present. Headings observedat present in the Carib-
United
States was similar
due to wetter
summers.
to the
Forest
the increased summer monsoon (Van Devender 1986). Most of ice-free
North
America
would
have
been excellent breeding habitat for migrants. The additional forest and grassland in the Southwestprobably increasedbreeding populations of passetinesand some shorebirdsin that areacomparedto presentconditions,while of warmer air abovea stabileanticycloneover winters would have been sufficiently cold to the icecap.In the Northwest(includinõAlaska) favor migrants over residents.Loss of coastal the North Central, and the Northeast, summers and wetland habitatsunder the ice cap(Fig. lB) were drier than at present. Due to a strenõth- may have been equaled by a gain in similar ened summer monsoon, the southern United habitat along the exposedcontinental shelf, Stateswere siõnificantly wetter than present. aroundglaciallakes,and along the multitudes The averaõestormtrackdipped southwardfrom of rivers and streamsdraining the melting ice western Canada to well below the Great Lakes cap. Speciescurrently nesting in tundra may and then ascendednorthwardalonõthe eastern havefound structurallysimilaralthoughwarmUnited Statescoastand out to sea.This pattern er habitats in the Southwest and near the meltprovided strohõ northwest winds west of cold ing ice. Tundra and coniferous forest habitats fronts over most of the continent (Fiõ. 215). were probably reduced compared to present Weaker winds were found in the Southwest and conditions,but there may have been a slight winds durinõ miõration were more northerly increase in mixed forest and deciduous forest in the south-centralUnited Statesthan alonõ habitatscomparedto the present.A generalshift the storm track (Fiõ. 215).Thus, winds over the in breeding habitat to the south and west is southern United Statesõenerally were lessfa- likely, due to increasedmoisture in the Southvorablethan ! 8 kya.The relativelysmallicecap west and elimination of habitat in the Northeast reduced sea level about 10 m below present under the ice cap. (Kutzbach 1987). This probably resulted in an The breeding area of the hypothetical miincreasein laõoonand wetland habitatin areas grant in Figure 2B is similar to that for present of continental shelf that were exposed(Emery conditionsexcept for areas of permanent ice. 1967).The meltinõ ice createdextensiveinland The southern boundary extendsfurther south lakessuchaslakeAgassiz(Fig. 2B).Winter tem- than at presentin the easternUnited Statesdue peratureswere similar to thoseat present. to greaterpresenceof confersin that area9 kya. Deciduous forests dominated eastern North At 9 kya, the headingsof migrants observed America, but the association of taxa differed from at present over Antigua would have brought the present (Jacobsonet al. 1987). A mixed co- birds from much the same areas of ice-free North niferous-hardwood forest extended north from Americaasthey do today (Fig. 2B).Birdsbreedthe GreatLakes.Prairie forbs,sedges,and grass- ing north of the ice 9 kya would have taken
north and with deciduous Pollen distributions and
trees in the south. macrofossils from
112
W•LL•MSANDWESS
bean could have been used from almost all North
American breeding areas9 kya. The retreat of ice in the easternportion of North America and the shift in the center of breeding distribution
[Auk,Vol. 113 DISCUSSION
Changesin climate during the most recent deglaciationof North Americaresultednot only to the north and east would have resulted in a in major changesin the area coveredby vegegreater percentageof birds moving over the tation, but also in the speciescomposition of Atlantic at 9 kya than during periods of full plant formations.During the pastmillion years, glaciation. presentconditionswere rare, representingnot 500 yearsago.--The present time is one of the norm, but the extreme developmentof bomaximum deglaciation.Boreal forest and tun- real forest and desert habitats.During the redra replacedmostof the remaining ice cap,pro- maining time, glaciation coveredmuch of the viding further habitatfor Neotropicalmigrants. northernbreedingareascurrentlyusedby NeoGrasslandsand desert expandedin the South- tropical migrants. Our analysisindicatesthat west and central United States due to a drier increasedvegetationin southernand coastalarclimate(Kutzbach1987,Thompson1988,Webb easmay have compensatedfor someof this loss 1988,Thompsonet al. 1993).The total areaavail- if northern speciescould competeeffectivelyin able for breeding as judged by our criteria, in- thesehabitats.The presentexpansionof tundra creasedrelative to 9 kya (Fig. 1A and B), but and taiga in North America is exceptional,and the removal of tree speciesfrom prairie areas, the abundanceof speciesexploitingthoseareas the retreat of forest areas in the Southwest and wasreducedsignificantlyduring periodsof glathe increasein desert in the Southwest,may ciation. The fauna of North America reacted to clihave reducedthe density of Neotropical passefinemigrantsthat couldbesupportedby these matechangein a variety of ways.Someanimals, habitats.Northern shorebirdhabitat appearsto such as southern reptiles, showed little change be near maximum at 500 ya (J•irvinenand V•iis- in distribution during the past20,000 years,in•inen 1978). Total populationsof boreal passer- steadthey adaptedto the changing conditions ines and shorebirdsat 500 ya may have been of the area(Van Devender1986).Although there near maximumfor the entire glacialcycle(giv- are few avian fossils from the Quaternary en that presenthabitatfor Neotropicalmigrants (Gauthreaux1980),it is probablethat birds, due hasbeengreatlyreducedby humanactivityfrom to their high mobility and observedcommon the levels at 500 ya). Winds, winter tempera- occurrenceof vagrancy, changed distribution tures,and averagestormtracksat presentare with the migration of plant speciesover North similar to thoseat 500 ya and 9 kya (Kutzbach America. This conclusion is consistent with the 1987). observedlow rate of Quaternary speciationin The expansionof habitat in the north and birdsascomparedwith mammals(Mengel 1964). east in the past 9,000 years and the reduction Gauthreaux(1980)and Huntley and Webb(1989) in habitat in the Southwest resulted in a northindependently concludedthat the movements eastwardshift in the centerof breedingdistri- of birds and of trees differed only in the time butions. The usefulness of fixed-heading ori- scaleof their migrations.As a descriptionof the entation for Neotropical migrants is more con- phenomenonthis is true, but birds are able to strained at 500 ya (and at present) than at any directtheir migrations,while treesarenot.Thus, other time in the glacial cycle.Birds departing the type of orientation systemusedby birds in from the eastern portion of the breeding migration may well have implications for groundson headingslessthan 155ø(illustrated changesin avian speciesdistribution. by the most easterntrack in Fig. 2A) would fly Flexibility in habitat selectionwould benefit too far east over the Atlantic to make landfall birds during periodsof major changesin plant in South America (Stoddard et al. 1983). Birds formations. Habitat selectionby birds includes departing the far western breeding range over bothinheritedpreferencesandreactionto proxthe Pacific on headings greater than 155ø also imate factors(Hutto 1985, Sherry and Holmes would fail to make landfall in South America. 1985).Selectionof both breeding and wintering Thus,althoughfixed-headingorientationwould sitesappearsdirectly influencedby competition be successfulfrom all of the breeding range, (Cody 1985,Cox 1985,Sherry and Holmes 1988, permissibleheadingsare severelyrestrictedin 1989,Holmes et aL 1989,Sherry 1990).Our analthe easternportionsand somewhatrestrictedin ysisagreeswith that of Cox (1985);habitat sethe far western portions of that range. lection in Neotropical migrant birds is more
January1996]
IceAgeBirdMigration andEcology
113
likely to be the result of interspecificcompetition than dependenceon specificplant species or plant associations. Suchspecificdependence would be difficult to reconcilewith relatively stableavian speciesin the faceof majorchanges
liams et al. 1977a,1981).Theselatter flights may constitutethe searchphase of orientation and,
in vegetation. Our estimatesof breeding habitats indicate that Neotropical migrants, even
tions).
rather than being the resultof errors,may be a specific adaptation to Quaternary conditions (Richardson1982, gives alternative interpreta-
Searchingfor suitablehabitat,necessitated by a fixed-headingorientation,may itselfhavebeen adaptiveduringperiodsof rapidclimatechange. greatly in abundanceas did the plant species Bothbirdsand mammalscouldtrackthe changes which they now exploit.During periodsof max- in plant distributionduring the Quaternary,yet imum glaciation, sufficient additional habitat most bird speciessurvived, while many mamwas available in presently unsuitable areasto mal speciesdid not (Cox 1985, Sutcliffe 1985). compensateat least in part for the loss of areas An effectivesystemfor discoveringnew, often coveredby ice. If this view is correct,the pres- distant habitatsand reevaluating old ones may ent observedreductionin Neotropicalmigrant have made a crucial difference. Current studies populations(Hagan and Johnston1992, Gill indicate that, for several speciesof passerine 1994) could be greater than that due to glacia- and shorebirdmigrants, successfuladults retion. turn to the samebreeding, stopover,and winFixed-heading orientation, coupled with diftering areas,while first-year birds and unsucferential reaction to weather as a stimulus for cessfuladult breedersdispersewidely (Sherry migration, appearswell suited to coping with and Holmes 1985,1989,Myers et al. 1987,JohnchangingQuaternaryecology.Although the ex- son et al. 1989). In most cases, birds that have act routes would vary from year to year, our found recentlyopened,underutilizedbreeding simulationsindicatethat sucha systemwould sites, wintering areas, or stopover sites could reliably direct birds over oceans,ice caps,and easilyrecoupthe costof a few hoursof flight. those presently restricted to boreal forest breeding areas, would have not have fluctuated as
continentalareas.Despitethe major shifts in breeding distribution shown in Figure 1, the southeastheadingsshown in Figure 2 continued to bring the great majority of migrants to Neotropicalwintering areasat all times tested. Somecueindicatingcessation of migrationmust be specified as well as direction. If this is distance or time as suggestedby Gwinner (1986), then this parameterwould have to changefor birds making longer flights. If the cue were instead a function of latitude, as would be the case for the elevation of the sun or stars, or
Present rather than past conditions appear to
posethe greateststresson a fixed-headingorientation system in our simulations. Considering that for 80 to 90%of the past900,000years North America has been in a stateof glaciation (Van Devender 1986, Ruddiman and Raytoo 1988), present conditionsmight be considered to be a rare period in avian evolution, posing a relatively brief stresson a system largely evolved
under
different
conditions.
Our
sim-
ulations suggestthat constant-headingorientation with a mean heading of 155øcould have evolved during periods of full glaciation. At that time it would have guided flights from breedingareasin southernNorth Americaover
inclination of the geomagnetic field (see Wiltschkoand Wiltschko1992),the systemcould function at all times in the Quaternarywithout change.Ratherthan directingbirds to a specific the Caribbean Sea and Gulf of Mexico to winlocation, this orientation systemwould bring tering grounds in the Caribbean Islands, Centhem to a large general area more than 100 km tral America, and South America. As the ice in diameter (Williams 1991). Birds would then retreated, the same orientation system would have to search within this area for suitable habdirect birds on longer and longer over-water itat (Williams and Williams 1978, 1990, Wiltschflights. During subsequentglaciations,the sysko and Wiltschko 1978, 1991). tem alsowould serveto crossthe ice capsthemRadar observationsof Neotropical migration selves,asis currently observedover Greenland revealboth large-scale broadfrontmigrationsas (Alerstam et al. 1986). The over-water and overpredicted for fixed-heading orientation and ice flights would result in energy savingsand (usuallyunder other weather conditions)flights reduced risk of predation comparedto flights of large numbers of birds in directions inap- around the ice capsor around seas(Williams et propriatefor the seasonalmigration (Drury and al. 1977a, Johnson and Herter 1990, Williams Keith 1962, Gauthreaux and Able 1970, Wil-
and Williams 1990)
114
WILLIAMS MaDWEBB
Orientation by means of a fixed southeast
heading is not the only systemused by Neotropical migrants (in the broad sense;Richardson 1991). Birds moving south to southwest through North America to Central America would haveto usesouthto southwestheadings, and Richardson(1979) observedlarge numbers of shorebirdsdepartingthe CanadianMaritime coaston headingstoo far eastto allow successful flight to South America without a change in orientation. Laboratory experimentswith orientation cages indicate multiple vectors consistingof fixed heading and distancesegments for Palearctic-African and Australian migrations (Wiltschko and Wiltschko 1991, Munro et
al. 1993).Geneticanalysesof thesevectorsshow that avian populationscan shift both direction and distancesignificantlyin only a few generations (Berthold 1990). These experiments suggestan alternative to our hypothesisof a relatively fixed, general heading; bird species could follow the displacementof specificplant speciesor speciesassociationsor specificinvertebrate fauna (Hutton 1984). In this view, mi-
gration is accomplishedby route-specificinstructions,and the presentmigratory routesof birds reflectthe pastaswell as presentlocation of the criticalhabitatsfor that bird species.Each successive stageof glaciationwould require alteringthe geneticsystemfor the migratoryroute or at leastthe addition of a new segmentto the encoded route. Some anomalies in present migration routescould be explainedby the persistenceof all or portionsof pastmigration routes in the species'geneticcode(Rabol 1978,Sharpe 1978,Gill 1994).Our analysisof North Americanvegetationhistory,the probablebenefitsof lesspreciseorientation systems,and the demonstrationof the efficacyof a simplefixed-heading systemsuggestthat suchroute-based,population-specificorientation systemsshould be rare in Neotropicalmigrants(asdefined for our simulations).This conclusionis supportedby the circular normal distributions of headings around
a southeast
mean
observed
for
Neo-
tropical migrantsover the western North Atlantic Ocean (Williams 1985). The relatively broadspreadof headingsaroundthis mean(SD > 25ø)is consistentwith birdsfacinguncertain conditionsin both winter and breeding areas. We suggestthat a more preciseorientation system (including one with lessvariation in fixed headings)would not benefit migrants. If it is true that a generalized orientation sys-
[Auk,Vol. 113
tern, such as fixed-headingorientation, allows a speciesto trackhabitatchangesover thousands of years,we still must accountfor its selective advantageduring the reproductivelifetime of a bird. Behaviorsthat are beneficial to a species are not favored in evolution unlessthey are also beneficial to the individual (Krebs and Davies
1987). A generalized, nonspecificorientation systemcoupled with extensive habitat search mustalsohave raisedthe averagemigrant'sfitnessabove that of a systemwhich reliably returned it to its natal location and to specific stopoverand wintering areas.A general characteristicof Quaternary ecology is that local habitatschangerelatively rapidly both in space and in time. Fires, floods, storms and severe
seasonal weathercreatea fluctuatingpatchwork of speciesassociationsand successionalstages. Raw datafrom pollen-coresitesshowthat plant associationsconsistedof a matrix of species changingirregularlyfrom siteto site.Only when many sitesare averageddo we find continuous variation over large areas(Baker1983,Jacobson et al. 1987). Palynologicalanalysison a highresolutiontime scaleshowsmany rapid changes in speciescomposition at a single site, some changesbeing asrapid as historicdeforestation from intensive, widespread lurebering (Jacobson et al. 1987). Records of lake or river sedi-
ments,shoreline positions,water temperatures, snowfall amounts, and volcanic or organic ash depositsall show variations at the finest discernible time intervals that are comparableto many of the average variations at intervals of hundredsor thousandsof years(Kutzbach1981, Street-Perrott
and Smith
1983, Kutzbach
and
Street~Perrott1985, Jacobsonet al. 1987, Engstrom et al. 1990, Gear and Huntley 1991, Koltermann and Gorelick 1992, Wang and Lewis 1992). Climate simulations and studies of Neo-
tropicalareassuggestsimilar variation for wintering areasof migrants(COHMAP 1988).South America and the Amazon basin appear to have supportedunstableassociations of speciesswept at intervals by fire and flood as recently as a few thousandyearsago (Meggers 1975, Colinvaux 1987). Under such conditions, it appears that the same factors
of unstable
climate
and
competitionthat favor habitat searchat present would also have been effective in increasing fitnessin the past. An average individual shift of as little as 0.5 km per year is sufficient to accountfor a population shift of 10,000km in 20,000 years. We conclude that fixed-heading
January 1996]
IceAgeBirdMigration andEcology
115
and migration of thrushes:A telemetry study. orientationandsearchwithin a goalareawould Illinois Nat. Hist. Surv. Bull. 33:297-330. benefit avian populationsduring the lifetime CODY, M. L. (ED.). 1985. Habitat selection in birds. of an individual and would alsoallow a species Academic Press, Orlando, Florida. to track large-scalevegetation changesover COHMAP MEMBERS.1988. Climatic changesof the thousandsof years. last 18,000 years:Observationsand model simulations.
Science 241:1043-1052.
COLINV^UX,P. 1987. Amazon diversity in light of the paleoecologicalrecord.Quat. Sci. Rev. 6:93-
ACKNOWLEDGMENTS
We thank: J. Kutzbachand P. Behling for climate-
114.
model simulations; and S. Johnson, R. Merz, W. J. Richardson,J. Weiner, J. Williams, and W. Wiltschko
Cox, G.W. 1985. The evolution of avian migration systemsbetween temperateand tropical regions
for their commentson drafts of this paper. This researchwassupportedby NSF grant ATM 87-13981to Brown University and by the Swarthmore College Faculty ResearchFund.
DENTON,G. H., AND T. J. HUGHES. 1981. The last
of the New
World.
Am. Nat.
126:451-474.
greatice sheets.Wiley Interscience,New York. DRURY,W. H., AND J. A. KEITH. 1962. Radar studies
of songbirdmigration in coastalNew England. Ibis 104:449-489.
LITERATURE CITED
EASTWOOD, E. 1967. Radar ornithology. Methuen, ABLE,K. P. 1970. A radar study of the altitude of nocturnalpasserinemigration. J. Field Ornithol. 41:282-290.
ABLE, K. P. 1980. Mechanisms of orientation, navi-
gation, and homing. Pages283-373 in Animal migration, orientation, and navigation (S. A. Gauthreaux, Jr., Ed). Academic Press, New York.
ALERSTAM, T. 1981. The courseand timing of bird migration. Pages9-54 in Animal migration (D. J. Aidley, Ed.). CambridgeUniv. Press,Cambridge. ALERSTAM,T., C. HJORT,G. HOGSTEDT,P. E. JONSSON,
J. KARLSSON, ANDB. LARSSON.1986. Spring migration of birds acrossthe Greenland Inlandice. Medd. Gronland, BioScience 21:3-38. AUBRY, D. G., K. O. EMERY, AND E. UCHUPI.
London.
EMERY, K.O. 1967. Estuariesand lagoonsin relation to continentalshelves.Pages9-11 in Estuaries(G. Lauff, Ed.). American Association for the Ad-
vancementof Science,Washington,D.C. EMERY,K. O., A. S. MERRILL,AND E. R. M. DRUFFEL.
1988. Changedlate Quaternarymarine environments on Atlantic continental shelf and upper slope. Quat. Res.30:251-269. EMERY,K. O., R. I. WIGLEY,A. S. BARTLETT, R. MEYER,
ANDE. S. BARGHOORN. 1967. Freshwaterpeat on the continental
shelf. Science
158:1301-1307.
ENGSTROM,D. R., B.C. S. HANSEN, AND H. E. WRIGHT,
JR. 1990. A possibleyounger Dryas record in 1988.
Changing coastallevels of South America and the Caribbean region from tide-gauge records. Tectonophysics154:269-284. BAKER, R. G. 1983. Holocenevegetationalhistory of the westernUnited States.Pages109-127 in Late Quaternaryenvironmentsof the U.S. Vol. 2, The Holocene (H. E. Wright, Ed.). Univ. Minnesota Press,Minneapolis. BARNOSKY, C. W., P. M. ANDERSON, AND P. J. BARWLEIN
southeastern
Alaska. Science 250:1383-1385.
GAUTHREAUX, S. A., JR. 1980. The influencesof longterm and short-termclimaticchangeson the dispersal and migration of organisms.Pages 103174 in Animal migration, orientation and navigation(S.A. Gauthreaux,Jr.,Ed.).AcademicPress, New
York.
GAUTHREAUX, S. A., JR. 1982. The ecology and evolution of avian migration systems.Avian Biol. 6:93-198.
1987. The northwestern U.S. during deglaciation: Vegetationalhistory and paleoclimaticimplications.Pages289-321 in North America and adjacentoceansduring the last deglaciation(W. F. Ruddimanand H. E. Wright, Eds.).The geology of North America,vol. K-3. GeologicalSocietyof
GAUTHREAUX, S. A., JR.,AND K. P. ABLE. 1970. Wind
America, Boulder, Colorado.
GILL,F. B. 1994. Ornithology, 2nd ed. W. H. Free-
and the direction of nocturnal songbird migration. Nature
228:476-477.
GEAR,A. J., AND B. HUNTLEY.1991. Rapid changes in the range limits of ScotsPine 4000 yearsago. Science 251:544-547.
man, New York.
BERTHOLD, P. 1988. The control of migration in Europeanwarblers. Pages215-249 in Acta XIX CongressusInternationalis Ornithologici (H. Ouel-
GWlNNER,E. 1986. Circannual rhythms. Springer Verlag, Heidelberg.
let, Ed.). Ottawa, Ontario 1986.National Museum
HAGAN, J. M., III, AND D. W. JOHNSTON(Eds.). 1992.
of Natural Science, Ottawa.
BERTHOLD, P. 1990. Geneticsof migration.Pages269283 in Bird migration: Physiologyand ecophysiology (E. Gwinner, Ed.).SpringerVerlag, Berlin. COCHRAN,W. W., AND C. G. KIos. 1985. Wind drift
Ecologyand conservationof Neotropicalmigrant landbirds. Smithsonian Institution
Press, Wash-
ington, D.C. HOLMES,R. T., T. W. SHERRY,AND L. RFaTSM^. 1989.
Population structure,territoriality and overwin-
116
WirErams ANDWEBB
[Auk,Vol. 113
ter survival of two migrant warbler speciesin
model
Jamaica. Condor 91:545-561.
in tropical lowland South America. Biotropica
HUNTLEY,B., ANDT. WEnBIII. 1989. Migration: Species' responseto climatic variations causedby changesin the earth'sorbit. J. Biogeog.16:5-19. HUTTO,R. L. 1985. Habitat selectionby nonbreeding, migratory land birds. Pages455-475 in Habitat selectionin birds (M. L. Cody,Ed.).Academic Press, Orlando, Florida.
HUTTON,E.J. 1984. The origin of migration of some bird species.Redstart51:99-102. IM•PaE,J., ^ND K. P. IMBRIE. 1979. Ice ages:Solving the mystery.Macmillan, London. J^CO•SON,G. L. J., T. WEn• III, ^ND E. C. GRIMM. 1987.
Patternsand rates of vegetation changeduring the deglaciationof easternNorth America.Pages 277-288 in North America and adjacentoceans during the lastdeglaciation(W. F. Ruddiman and H. E. Wright, Eds.).The geologyof North America,vol. K-3. GeologicalSocietyof America,Boulder, Colorado.
of diversification
to cultural
distributions
7:141-161.
MENGEL,R.M. 1964. The probable history of species formation in some northern wood warblers (Pa-
rulidae). Living Bird 3:9-43. MENGEL,R.M. 1970. North American central plains
as an isolating agent in bird speciation.Pages 279-339
in Pleistocene
and Recent environments
of the central Great Plains (W. Dort and J. K.
Jones,Jr.,Eds.).University Pressof Kansas,Lawrence.
MOREAU, R. E.
1972.
The Palaearctic-African
bird
migration systems.AcademicPress,New York. MUNROE, U., W. WILTSCHKO, AND H. A. FORD. 1993.
Changes in the migratory direction of Yellowfaced HoneyeatersLichenostomus chrysops (Meliphagidae)during autumn migration. Emu 93:5962.
MYERS, J. P., R. I. G. MORRISON, P. ANTAS, B. A. HARRINGTON, T. E. LOVEJOY,M. SALLABERRY, S. E.
SENNER, AND A. TARAK. 1987. Conservation J.•RVINEN, O., ANDR. A.V.•IS.•NEN. 1978. Ecological zoogeographyof northern Europeanwaders,or strategyfor migratoryspecies.Am. Sci.75:18-26. J. T., R. S. WE•, AND T. WE• III. 1992. why do so many waders breed in the North? OvERPECK, Oikos 30:496-507. Mapping eastern North American vegetation JOHNSON,O. W., M. L. MORTON,P. L. BRUNERAND P. change over the past 18,000 years: No-analogs M. JOHNSON.1989. Fat cyclicity, predicted miand the future. Geology20:1071-1074. RESEARCH WORKING GROUP OF PARTNERS IN FLIGHT. gratory flight ranges,and featuresof wintering behaviorin PacificGolden-Plovers(Pluvialis ful1992. Justwhat is a Neotropical migrant?Partva). Condor 91:156-177. ners in Flight Newsletter 2(1):29-30. JOHNSON,S. R., AND D. R. HERTER. 1990. Bird miPELTIF, R, W.R. 1987. Glacial isostasy,mantle viscosgration in the Arctic: A review. Pages 22-43 in ity, and Pleistoceneclimatic change.Pages155Bird migration: Physiologyand ecophysiology 182 in North Americaand adjacentoceansduring (E. Gwinner, Ed.). Springer Verlag, Berlin. the last deglaciation(W. F. Ruddiman and H. E. KOLTERMANN,C. E., AND S. M. GORELICK. 1992. PaWright, Eds.).The geologyof North America,vol. leoclimatesignature in terrestrial flood deposits. K-3. GeologicalSocietyof America,Boulder,ColScience 256:1775-1782.
KRESS,J. R., AND N. B. DAVIES. 1987. An introduction
to behavioral ecology. Sinauer Associates,Sunderland, Massachusetts.
KUTZl•ACH, J. E. 1981. Monsoonclimate of the early Holocene: Climate experiment with the earth's orbital parametersfor 9000 years ago. Science 214:59-61.
KUTZBACH, J.E. 1987. North America. Pages425-445 in North Americaand adjacentoceansduring the last deglaciation (W. F. Ruddiman and H. E. Wright, Eds.).The geologyof North America,vol. K-3. GeologicalSocietyof America,Boulder,Colorado.
KUTZBACH,J. E., AND F. A. STREET-PERROTT. 1985. Mil-
ankovitch forcing of fluctuations in the level of tropical lakes from 18 to 0 kyr BP. Nature 317: 130-134.
McCLINTOCK, C. P., m. C. WILLIAMS, AND J. M. TEAL.
1978. Autumnal bird migration observed from shipsin the western North Atlantic Ocean.BirdBanding49:262-275. MEGGEE, S, B. J. 1975. Application of the biological
orado.
RABOL, J. 1978. One-directionorientationversusgoal areanavigationin migratorybirds.Oikos30:216223.
RICHARDSON, W. J. 1972. Autumn migration and weather in easternCanada:A radar study. Am. Birds 26:10-17.
RICHARDSON, W.J. 1978. Timing and amount of bird migrationin relationto weather:A review. Oikos 30:224-272.
RICHARDSON, W. J. 1979. Southeastward shorebird
migration over Nova Scotiaand New Brunswick in autumn: A radar study. Can. J. Zool. 57:107124.
RICHARDSON, W. J. 1980. Autumn landbird migration over the western
Atlantic
Ocean as evident
from radar. Pages501-506 in CongressusInternationalisOrnithologici (R. N6hring, Ed.).Berlin 1978. DeutscheOrnithologen-Gesellschaft,Berlin.
RICHARDSON, W.J., 1982. Northeastward reverse mi-
gration of birds over Nova Scotia, Canada, in autumn.
Behav. Ecol. Sociobiol.
10:193-206.
January1996]
IceAgeBirdMigration andEcology
RICHARDSON, W.J. 1990. Timing of bird migration in relationto weather:Updatedreview. Pages78101 in Bird migration:Physiologyand ecophysiology (E. Gwinner, Ed.). SpringerVerlag, Berlin. RICHARDSON, W.J.
1991. Wind and orientation of
migratingbirds:A review. Pages226-249 in Orientation in birds (P. Berthold, Ed.). Birkh•iuser
Verlag, Basel. RICHARDSON,W. J., AND W. W. GUNN. 1971. Radar observations
of bird movements
in east-central
Alberta. Can. Wildl. Serv. Rep. Set. 14:35-68. RUDDIMAN, W. F., AND M. E. R•¾MO. 1988. Northern
Hemisphereclimate regimesduring the past 3 Ma: Possible tectonic connections.
Phil. Trans. R.
Soc. Lond. B 313:409-428.
SHARPE, R.S. 1978. The originsof springmigratory staging by Sandhill Cranes and White-fronted Geese. Trans. Nebr. Acad. Sci. 6:141-144.
SHERRY, T. W. 1990. When are birds dietarily specialized?Distinguishing ecologicalfrom evolutionary approaches.Stud.Avian Biol. 13:337-352. SHERRY, T. W., AND R. T. HOLMES. 1985. Dispersion patternsand habitat responsesof birds in northern hardwood forests.Pages283-309 in Habitat selectionin birds (M. L. Cody, Ed.). Academic Press, Orlando, Florida. SH•R¾, T. W., AND R. T. HOLMES. 1988. Habitat se-
lection by breeding American Redstartsin responseto a dominantcompetitor,the LeastFlycatcher. Auk 105:350-364.
SHERRY, t. W., .•JqDR. T. HOLMES.1989. Age specific socialdominanceaffectshabitatuseby breeding AmericanRedstarts(Setophaga ruticilla):A removal experiment.Behav.Ecol.Sociobiol.25:327-333. STODDARD, P. K., J. E. MARSDEN,AND T. C. WILLIAMS.
1983. Computersimulationof autumnalbird migration over the western North Atlantic. Anim. Behav. 31:173-180.
117
nities: The late Pleistocenepackratmidden record. Pages285-299 in Community ecology(J. C. Diamond and T. Case, Eds.). Harper and Row, New
York.
WALLRAFF, H. G. 1991. Conceptual approachesto avian navigationsystems.Pages129-165 in Orientation in birds (P. Berthold, Ed.). Birkh•iuser Verlag, Basel. WANG, K., AND T. LEw•s. 1992. Geothermal evidence
from Canadafor a cold period before recent climatic warming. Science256:1003-1005. WEss,T., III. 1988. EasternNorth America. Pages
385-414in Vegetationhistory(B. Huntley and T. Webb III, Eds.).Kluwer, Dordrecht, Netherlands. WESS,T., III, P. J. BARTLEIN, AND J. E. KUTZBACH. 1987.
Climatic changein easternNorth America during the past 18,000years:Comparisonsof pollen data with model results.Pages447-462 in North Americaand adjacentoceansduring the last deglaciation (W. F. Ruddiman and H. E. Wright, Eds.). The geology of North America, vol. K-3. GeologicalSocietyof America, Boulder, Colorado.
WILLIAMS, T.C. 1985. Autumnal bird migration over the windward
Caribbean
Islands. Auk
102:163-
167.
WILLIAMS, T.C. 1991. Constantcompassorientation for North Americanautumnal migrants.J. Field. Ornithol.
62:218-225.
WILLIAMS, T. C., P. BERKELEY, AND V. HARRIS. 1977b.
Autumnalbird migrationover Miami studiedby radar:A possibletest of the wind drift hypothesis.Bird-Banding48:1-10. WILLIAMS,T. C., J. E. MARSDEN,T. L. LLOYD-EVANS, V.
KRAUTHAMER, ANDH.KRAUT•mMER. 1981. Spring migrationstudiedby mist-netting,ceilometer,and radar. J. Field Ornithol. 52:177-190. WILLIAMS, T. C., J. M. WILLIAMS, L. C. IRELAND, AND
S'rRE•r-Pm•ROrr, F. A., AND G.I. SMITH. 1983. Pluvial
J.M. TEAL. 1977a. Autumnal bird migration over
lakes of the western United States.Pages190212 in LateQuaternaryenvironmentsof the U.S. Vol. 2, The Holocene(H. E. Wright, Ed.). Univ. Minnesota.Press,Minneapolis. SV'rCLIFFE, A.J. 1985. On the track of the ice age mammals.Harvard Univ. Press,Cambridge,Mas-
the western North 251-267.
sachusetts.
Atlantic
Ocean. Am. Birds 31:
WILLIAMS,T. C., AND J. M. WILLIAMS. 1978. Orien-
tation of transatlanticmigrants.Pages239-251 in Animal migration, navigation,and homing (K. Schmidt-Koenigand W. T. Keeton, Eds.). Proceedingsin the Life Sciences,Springer-Verlag,
Berlin. THOMPSON, R.S. 1988. WesternNorth America.Pages WILLIAMS,T. C., ANDJ. M. WILLIAMS. 1990. The ori415-458 in Vegetationhistory (B. Huntley and T. Webb III, Eds.).Kluwer, Dordrecht, Netherlands entation of transoceanicmigrants.Pages9-21 in THOMPSON,R. S., C. WHITLOCK,P. J. BARTLEIN,S. P. Bird migration: Physiology and ecophysiology HARRISON, AND W. G. SPAULDING. 1993. Climatic (E. Gwinner, Ed.). Springer Verlag, Berlin. changesin the westernUnited Statessince18,000 WILTSCHKO, R. 1992. Das Verhalten verfrachteter yr B.P.Pages468-513 in Global climatessincethe V/Sgel.Vogelwarte 36:249-310. WILTSCHKO, W., AND R. WILTSCHKO. 1978. A theolastglacialmaximum(H. E. Wright, Jr.,J.E. Kutzbach, T. Webb III, W. F. Ruddiman, F. A. Streetreticalmodel for migratoryorientationand hom-
Pertort, and P. J. Bartlein, Eds.).Univ. Minnesota Press,Minneapolis. VAN DEVENDER,T. R.
1986. Climatic cadences and
the compositionof Chihuahuan Desert commu-
ing in birds. Oikos 30:177-187. WIVrSCHKO, W., ANDR. WIL'rSCHKO.1988. Magnetic orientation
in birds. Curt.
Ornithol.
5:67-121.
WIL'rSCHKO, W., AND R. WIL'rSCHKO.1991. Magnetic
118
WILLIAMS ANDWI•B
orientation and celestialcuesin migratory orientation. Pages16-37 in Orientation in birds (P. Berthold,Ed.). BirkhauserVerlag, Basel. WILTSCHKO, W., ANDR. WILTSCHKO.1992. Migratory
[Auk,Vol. 113
orientation: Magnetic compass orientation of Garden Warblers (Sylviaborin)after a simulated crossingof the magnetic equator. Ethology 91: 70-74.
