AMER. ZOOL., 31:187-204 (1991)
The Flight Behavior of Migrating Birds in Changing Wind Fields:
Radar and Visual Analyses'
SIDNEY A. GAUTHREAUX, JR.
Department of Biological Sciences, Clemson University, Clemson, South Carolina 29634-1903
SYNOPSIS. This paper examines the influence of atmospheric structure and motion (principally winds aloft) on the flight behavior and altitudinal distribution of migrating songbirds. Bird migration data that I gathered using surveillance radars operated by the United
States National Weather Service and the Federal Aviation Administration and a vertically
directed fixed-beam marine radar mounted on a mobile laboratory are analyzed in relation
to winds aloft. Migrating birds appear to fly at altitudes where winds will minimize the
cost of transport and assist movements in seasonally appropriate directions. When migratory flights occur at altitudes that are higher than usual, a significant correlation exists
between the altitude of densest migration and the altitude of most favorable wind. Lower
altitudes may be favored over slightly more favorable winds at much higher altitudes.
Radar data on the flight behavior of migrating birds in the vicinity of frontal systems is
also examined. The flight strategies of migrants (fly over the front, change the direction
of flight, or land and terminate theflight)differ depending on season and the "thickness"
of the front. Recent migration studies that are related to atmospheric structure and motion
are summarized and related to atmospheric processes operating simultaneously at vastly
different spatial and temporal scales.
MEASUREMENTS AND TECHNIQUES
INTRODUCTION
The movements of aerial migrants
undoubtedly have been greatly influenced
by the structure and motion of the atmosphere, but not until zoologists began using
radiotelemetry and radar in their migration studies was it possible to study migrant-atmospheric interactions in great
detail. Important information on the flight
strategies of individual birds has been collected with the aid of radiotelemetry (e.g.,
Cochran etal, 1967; Cochran, 1972;Cochran and Kjos, 1985) and tracking radars
(e.g., Bruderer and Steidinger, 1972; Able,
1977; Alerstam, 1985; Larkin and Thompson, 1980), but considerably more data have
been gathered with surveillance and fixedbeam radar systems since the publication
of Eastwood's (1967) pioneering volume
Radar Ornithology. The migration of insects
in relation to atmospheric processes has
been studied not only with surveillance
weather radars (e.g., Richter et al., 1973)
but also with radar systems developed specifically for studying insect migration and
dispersal (e.g., Schaefer, 1976, 1979; Sparks
etal, 1985).
The data on bird migration-atmospheric
interactions that I present in this paper have
1
From the symposium on Recent Developments in the been gathered with three different radar
Study of Animal Migration presented at the annual
meeting of the American Society of Zoologists, 27— systems: weather surveillance (WSR-57),
airport surveillance (ASR-7), and marine
30 December 1988, at San Francisco, California.
The importance of atmospheric structure and motion cannot be overstated when
considering the selective pressures that
shape the flight behavior of migrants
through the atmosphere. Much of the
information gathered on the influence of
atmospheric structure and motion on insect
migration has been recently summarized
by Drake and Farrow (1988) and the
importance of atmospheric structure on the
diel timing and altitude of bird migration
has been reviewed by Kerlinger and Moore
(1989). In the paper that follows I will not
reiterate the material covered in these
papers but instead emphasize additional
published and unpublished information on
the flight behavior of migrating birds in
response to changing wind fields at different spatial and temporal scales. Although
I will focus on data collected during my
radar studies of bird migration on the
northern coast of the Gulf of Mexico and
elsewhere in the Southeast, pertinent published information on migratory movements through the atmosphere will be discussed when appropriate.
187
188
SIDNEY A. GAUTHREAUX, JR.
surveillance (LN 66 and Raytheon 3400).
Details of the use of the WSR-57 and the
ASR-7 radars to detect, monitor, and
quantify bird migration can be found in
Gauthreaux (19806), and details of the use
of the LN 66 and Raytheon 3400 radars
as vertical fixed-beam units can be found
in Gauthreaux (1985).
and a 3-cm wavelength (X-band, 9,3459,405 MHz). The Raytheon 3400 radar has
replaced the LN 66 and it has a peak power
of 5 kw and the same wavelength as the
LN 66. The fixed antenna is mounted
directly on the transmitter/receiver unit
and points straight overhead.
Visual confirmation of radar echoes
WSR-57 radar
The WSR-57, operated by the U.S.
National Weather Service, has 500 kw peak
power, a 10-cm wavelength (S-band, 2,7002,900 MHz), and a 3.7 m parabolic antenna
that focuses a 2° beam. The angle of
antenna tilt can be changed, and in the
surveillance mode the antenna rotates once
every 20 sec. Once stopped the antenna
can be used to scan vertically in the rangeheight mode. For most observations the
radar range displayed was 25 nautical
miles—nm (46 km), but occasional observations were made at a range of 125 nm
(231 km).
Whenever possible I confirmed visually
the identity of the sources of radar echoes
displayed on the radar screen (PPl-plan
position indicator) using three different
methods: moon-watching or ceilometer
beam observations at night and vertical telescope or binocular observations during the
day. The details of moon-watching can be
found in Lowery (1951) and Nisbet (1959),
and the methods of using binoculars, telescope, or image intensifier to observe
migrants aloft in a narrow beam of light
are detailed in Gauthreaux (1969, 19806,
1985), and Able and Gauthreaux (1975).
Dates and locations
ASR-7 radar
The data from WSR-57 radar observaThe ASR-7, operated by the U.S. Fed- tions included in this paper were gathered
eral Aviation Administration, has 425 kw at Athens, Georgia (fall 1968-spring 1971);
peak power, a 10-cm wavelength (S-band, Charleston, South Carolina (fall 1970-fall
2,700-2,900 MHz), and a slotted dish 1975); and Lake Charles and New Orleans,
antenna (2.7 m high and 5.2 m wide) with Louisiana (fall 1964-spring 1968). ASR-7
a vertical fan beam 1.5° horizontal and 5° observations were made at Greenville and
vertical (cosecant squared to 30°). The Charleston, South Carolina, and Lake
antenna rotates once every 4 sec. The radar Charles, Louisiana (fall 1971-spring 1983).
is equipped with a moving target indicator Most of the observations with the fixed(MTI) so that only moving targets are dis- beam marine radar were made at Lake
played in a typical "figure 8" or "hour- Charles, Louisiana (spring 1981, 1982,
glass" pattern. For most observations the 1983).
radar range displayed was 10 nm (18 km),
but occasionally the range was changed to Meteorological and climatic information
6 nm (11 km) and 20 nm (37 km).
I gathered local meteorological information
on wind, temperature, barometric
Marconi LAT 66 and
pressure, humidity, and cloud cover every
Raytheon 3400 marine radars
hour during migration watches from a
I modified these small marine surveil- nearby National Weather Service office.
lance radars by replacing the original "T- Winds aloft information were gathered by
bar" antenna with a parabolic dish antenna the National Weather Service at 00:00 and
(60 cm diameter) that has a 4° beam. The 12:00 GMT by radiosonde (radar tracked
modified radar was mounted on the roof balloon) and occasionally at 06:00 GMT by
of a mobile laboratory (see Gauthreaux, pibal (visually tracked balloon) near all
1985). The LX 66 radar was the first radar radar stations except Greenville, South
used, and it had a peak power of 10 kw Carolina. For analysis of the Greenville
MIGRATING BIRD FLIGHT BEHAVIOR
189
2286
2134
1981
1829
1676
1524
1372
1219
1067
914
762
tO
20
30
40
50
Number of Flocks
|
c
10
20
Cases
FIG. 1. Altitudinal distribution of arriving trans-Gulf migrations. A. Distribution of flocks during one measurement. B. Altitudinal distributions of peak densities of 92 trans-Gulf flights over Lake Charles and New
Orleans, Louisiana, combined from 1965 through 1967. C. New Orleans data alone (57 measurements). D.
Lake Charles data alone (35 measurements).
migration data I used the winds aloft data
from Athens, Georgia; Nashville, Tennessee; and Greensboro, North Carolina.
distribution is the basic unit in the other
three diagrams (Fig. 1B-D). In Figure 4B
all altitudinal data from both radar stations
over the three years are combined. The
data
are separated for New Orleans (Fig.
RESULTS AND DISCUSSION
4C) and Lake Charles (Fig. 4D). Arriving
Altitudinal distribution of arriving
trans-Gulf flights were on the average
trans-Gulf migrants
higher over New Orleans than over Lake
The day-to-day variance in the altitudi- Charles. This is likely associated with the
nal distributions of arriving trans-Gulf fact that the migrants rather consistently
flights on the northern coast of the Gulf flew above the convective cumulus clouds
of Mexico in spring can be considerable. that formed near the coastline and develThe variance in large part can be explained oped as they moved inland. Because New
by the dynamics of the vertical structure Orleans was farther inland the clouds
of the atmosphere over the area during this tended to be taller over New Orleans than
time of the year. The mean altitudes of over Lake Charles. Once the clouds develarriving trans-Gulf flights measured with oped beyond a height of 2,100 m the
WSR-57 radars at New Orleans and Lake migrants usually lowered their altitude.
Charles, Louisiana, at 06:00 and 18:00 CST The importance of the winds aloft to the
from mid-March through mid-May for altitudes of the trans-Gulf flights (Fig. 1)
three different years are plotted in Figure can be seen by examining Figure 2. The
1A-D. Figure 1A shows a plot of the alti- wind directions and speeds have been avertudinal distribution of mostly passerine bird aged for the same three springs for which
flocks for one case. The median of such a altitudinal measurements are available. The
190
SIDNEY A. GAUTHREAUX, JR.
FIG. 2. Wind directions and speeds around the Gulf of Mexico (three-year average—1965-1967 for 16
stations). Maps on left side show surface winds; maps on right side show winds at 900 millibar level (914 m).
A-B. March, C-D. April, E-F. May.
maps on the left side (Fig. 2A, C, and E)
show surface winds for the months of
March, April, and May, while the maps on
the right side (Fig. 2B, D, and F) show winds
at the 900-millibar level or approximately
915 m (3,000 ft) for the same three months.
Although surface flow during the spring
can hardly be considered favorable for the
191
MIGRATING BIRD FLIGHT BEHAVIOR
o
3000
2750
2500
2250
2000
1750"
15001250"
1000"
750"
500"
250"
0
160
180
200
220
240
Wind Direction (from)
10
12
14
16
18 20
Frequency
FIG. 3. Winds aloft and the altitudinal distribution of trans-Gulf migrants arriving on the northern coast of
the Gulf of Mexico in the spring of 1981. A. Winds aloft with speeds in knots (1 knot = 0.5 m/sec). B.
Frequency of migrants at various altitudes.
transport of trans-Gulf migrants from areas
south of the Gulf, the winds aloft are very
conducive to trans-Gulf flight.
During three additional spring seasons
in southwestern Louisiana I measured with
the fixed-beam marine radar the altitudinal distributions of passerines arriving from
over the Gulf of Mexico during the daylight hours. The winds aloft and the altitudinal distribution of migrants are plotted
for each of the three springs (Figs. 3, 4,
and 5). The mean altitude of the arriving
migrants (46 individuals) measured during
one flight in the spring of 1981 was 0.84
km (Fig. 3B) and at this altitude the wind
aloft was blowing from 200° at 16 kts or
8.2 m/sec (Fig. 3A). The mean altitude of
the arriving migrants (2,619 individuals)
measured during five flights in the spring
of 1982 was 0.98 km (Fig. 4B) and at this
altitude the mean wind aloft was blowing
from 158° at 20.6 kts or 10.6 m/sec (Fig.
4A). On these five occasions the mean wind
aloft did not shift to the west of south until
an altitude of 2.5 km. During the spring
of 1983 five flights were monitored (Fig.
5A-D). The mean altitude of the arriving
migrants (1,574 individuals) measured
during five flights in the spring of 1983
was 0.4 km. The low mean altitude of these
flights was strongly biased by the data collected on 7 May. On that day the mean
altitude of 1,356 individuals was 0.4 km
(Fig. 5B) and the wind aloft at this altitude
was 200-235° blowing at 8 kts or 4.1 m/sec
(Fig. 5A). When the data from 7 May are
excluded, the mean altitude of the flights
on the remaining four days is 0.88 km (Fig.
5D) and at this altitude the mean wind aloft
was blowing from 166° at 21.2 kts or 9.1
m/sec (Fig. 5C). The mean winds aloft for
the four spring flights in 1983 are given in
Figure 5C.
Distribution of nocturnal migrants aloft
The nighttime migration of most songbirds usually occurs at altitudes below 500
m above ground level, but occasionally most
of the migrants aloft are flying at much
higher altitudes. By concentrating on these
instances it is possible to understand the
atmospheric processes associated with such
deviations. As will be shown, by emphasizing the cases when migrating songbirds are
flying higher in the night sky than is normal, it is possible to demonstrate flight tactics in response to changing wind fields
aloft. The data in this section were gathered during radar and visual studies of bird
migration in Louisiana, Georgia, and South
Carolina, since the autumn of 1964. The
results of analysis of the data gathered with
the WSR-57 radar will be presented first
followed by the results obtained from the
192
3
SIDNEY A. GAUTHREAUX, JR.
3250
3000
2750"
3250"
2500"
2500-
2250"
2000"
17501500"
1250"
1000750"
500'20
250'12
0
140 160 180 200 220
3000-
B
2750225020001750150012501000750500250 1
0
Wind Direction (from)
50
100
150
200
250 300 350 400 450 500
Frequency
FIG. 4. Winds aloft and the altitudinal distribution of trans-Gulf migrants arriving on the northern coast of
the Gulf of Mexico in the spring of 1982. A. Winds aloft with speeds in knots (1 knot = 0.5 m/sec). B.
Frequency of migrants at various altitudes.
analysis of data gathered with the ASR-7
radar.
WSR-57 radar measurements. From the
autumn of 1964 through the autumn of
1975, 79 measurements of the altitudinal
distribution of nocturnal migrants aloft
were made on 70 nights in spring, and 39
altitudinal measurements were made on 35
nights in fall. Of the 79 altitudinal measurements made in spring, on 21 occasions
the densest migration occurred at an altitude of 450 m or higher (Table 1). Of the
39 measurements made in the fall, on 17
occasions the densest concentrations of
migrants aloft occurred at an altitude of
450 m or higher (Table 1). In an effort to
see if the variance in altitudinal distributions was correlated with wind patterns
aloft, I plotted the altitude of densest
migration and the altitude of most favorable wind for each case. The most favorable wind was denned as a wind blowing
toward the N-NE in spring or S-SW in fall
and occurring at the lowest possible altitude. If winds at all altitudes were blowing
in unfavorable directions (e.g., opposite the
normal seasonal direction of migration), the
most favorable wind was the slowest in
velocity.
Only the 38 cases (21 in spring and 17
in fall) when the densest concentrations of
songbirds were at 457 m or at higher altitudes (from Table 1) are plotted in Figure
6. The correlation between the altitude of
densest migration and the altitude of most
favorable wind is highly significant (r =
0.96, P < 0.0001). The relationship is quite
possibly better than indicated, for wind
conditions aloft can change from hour-tohour, and usually the radar measurements
were made 2 hr after the wind aloft measurements were made. In the Southeast the
winds at 305 m usually blow in directions
that are favorable for passerine migrants
to move to the north and northeast in
spring and to the south and southwest in
fall. In Figure 7A I have plotted the directions toward which the 305 m winds were
blowing on 47 of the 57 occasions in spring
when the altitude of greatest concentration of birds aloft occurred near 305 m.
The remaining ten cases will be discussed
with Figure 8. The resultant direction of
the winds is 37.6° (r = 0.75; angular deviation, s = 40°).
When the migrations occurred at altitudes of 475 m or higher the winds where
the birds were flying were quite similar in
direction to those recorded when the
migrations occurred at 305 m (Fig. 7B).
193
MIGRATING BIRD FLIGHT BEHAVIOR
4)
3
3250-1
30002750 "
2500"
2250"
2000"
1750 "
1500"
1250"
1000750500250"
0
160 180 200 220 240 260
3250
3000
2750
2500
2250
2000
1750
1500
1250
1000
750
500
250
0
B
100 200 300 400 500 600 700 800
Frequency
Wind Direction (from)
u
T3
3
3250"
3250"
3000275025002250"
2000"
17501500"
1250"
1000750"
500"
250"
0
30002750"
D
2500"
2250"
20001750"
15001250"
1000750"
500"
250-
0
150 160 170 180 190 200 210
0
5 10 15 20 25 30 35 40 45 50 55 60
Frequency
Wind Direction (from)
FIG. 5. Winds aloft and the altitudinal distribution of trans-Gulf migrants arriving on the northern coast of
the Gulf of Mexico in the spring of 1983. A. Winds aloft for 7 May (speed in knots). B. Altitudinal distribution
of migrants aloft on 7 May. C. Mean winds aloft for 6, 11, 12, and 13 May (speed in knots). D. Altitudinal
distribution of migrants aloft on 6, 11, 12, and 13 May.
For the 21 occasions in spring when the
altitudes of migration were 475 m or
higher, the resultant direction of the winds
is 41° (r = 0.75; s = 40.4°). Thus in spring
the winds at 305 m are usually favorable
for the northward progression of migrants,
but when the 305 m winds are unfavorable,
the birds often fly higher if favorable winds
can be found at higher altitudes.
Figure 7C shows the directions toward
which the 305 m winds were blowing on
18 of the 22 occasions in fall when the
greatest concentrations of migrants
occurred at 305 m (the remaining four cases
will be discussed with Fig. 8). The resultant
direction of the 305 m winds is 243° (r =
0.71; s = 44°). The open circles represent
TABLE 1. Altitude of greatest concentration of nocturnal
migrants aloft.
Number of observations
Altitude
Meters
Feet
Spring
Fall
152
305
457
610
762
914
1,219
1,372
1,524
1,676
1,829
2,134
2,286
500
1,000
1,500
2,000
2,500
3,000
4,000
4,500
5,000
5,500
6,000
7,000
7,500
1
57
3
6
2
—
2
3
1
1
2
1
—
79
22
2
3
1
1
1
—
3
2
2
1
1
39
194
SIDNEY A. GAUTHREAUX, JR.
2440
2135
•
2
1830
or
o
=
1525
** /
*
1220
/
°
Y = 58-6 • 0 - 9 8 7 (X)
915 •
•
•
»
0 93 (93%)
t = 22 9 p< 0 0 1
610 '
n = 38
305 "
305
610
915
1220
1525
1830
2135
ALTITUDE OF MOST FAVORABLE WINDS ALOFT
2440
(IN METERS)
FIG. 6. Correlation between the altitude of densest bird migration at night and the altitude of most favorable
winds aloft.
winds that are nearly calm (less than 3 km/
hr). In the fall when the altitudes of migrations were 475 m or higher, the winds
where the birds were flying were either
similar to the average direction at 305 m
or were nearly calm (Fig. 7D). The resultant direction of the winds associated with
the 17 flights at altitudes of 475 m or higher
is 208° (r = 0.35; s = 65°). When the nearly
calm winds (open circles) are eliminated,
the resultant direction of the winds is 200°
(r = 0.75; s = 40°). The more southerly
resultant direction of the higher altitude
winds is in keeping with the tendency of
winds to blow more to the east at greater
altitudes in the northern temperate zone.
On 10 occasions in spring and four in
fall the densest migration occurred at 305
m, winds at higher altitudes were more
favorable (Fig. 8). The migrants in each
case were flying in the same general direction as the 305 m winds. In Figure 8 the
arrows in the inner circles show the wind
directions at 305 m, and the arrows in the
outer circles show the direction of winds
at the altitudes indicated at the lower right
of each circular diagram. In most of the
spring cases (Fig. 8A-C, E, F, H, J) the
migrants are moving to the west and northwest, but winds blowing closer to north and
northeast are available at higher altitudes.
In Figure 8D, G, and I, the migrants are
flying toward the southwest with the wind
at 305 m (reverse migrating), but the winds
are only marginally more favorable at a
higher altitude in cases D and I. The case
illustrated in Figure 8G is interesting
because winds about 610 m higher were
quite favorable, however, the 305 m winds
for the previous three nights were quite
favorable for migration in a north-northeasterly direction, so that it is possible that
the birds involved in Figure 8G had destinations toward the southwest.
In the fall four cases were recorded when
migrants were flying with the 305 m winds
when at higher altitudes there were somewhat more favorable winds for flights to
the south (Fig. 8K-N). These cases suggest
that lower altitudes may be favored over
MIGRATING BIRD FLIGHT BEHAVIOR
195
surements with the WSR-57 radar would
be similar using an entirely different data
set, I analyzed a data set of altitudinal measurements gathered with the use of the
ASR-7 radar. Because the ASR radar has
a wide vertical beam, altitudinal measurements were less accurate than those made
with the WSR-57, but changes in the altitudinal distribution of migrants in the night
FALL
sky could be measured by recording the
shifts in the distribution (density) of echoes
on the radar screen as a function of range
(Fig. 9). From the autumn of 1971 through
the spring of 1983, 184 measurements were
made on 184 nights in spring, and 13 measurements were made on 13 nights in fall.
Fic. 7. Wind directions at the altitudes of densest The altitudinal distribution of the meabird migration. A. 47 occasions when peak concen- surements are given in Figure 10. Of the
trations of migrants occurred at or below 305 m in
197 measurements 38 (19%) show flights
spring. B. 21 occasions when peak occurred at 475 m at altitudes greater than 500 m. Thirty-six
and above. C. 18 occasions when peak occurred at
305 m and lower in fall. D. 17 occasions when peak (20%) of the 184 measurements in spring
occurred at 475 m and above. Open circles represent and 2 (15%) of the 13 measurements in fall
wind velocities less than 1 m/sec.
are above an altitude of 500 m.
When the altitudes of greatest concenslightly more favorable winds at much tration of migrants aloft (above an altitude
higher altitudes. In Figure 8L the migrants of 500 m, 38 cases) are plotted against the
would have had to climb 1,890 m to take altitude of most favorable wind aloft for
advantage of a wind blowing 65° closer to the same night (Fig. 11), the resulting correlation is significant (r = 0.57,P < 0.0002)
south.
ASR-7 radar measurements. In an effort to but clearly not as good as that found in the
see if the findings based on altitudinal mea- WSR-57 analysis. The coarseness of altiSPRING
1525m
1220m
FIG. 8. Cases when the densest migration occurred at 305 m and below but winds at higher altitudes were
more favorable. The altitude with the most favorable wind is given in the lower right of each circular diagram
and the wind speeds are given in knots for both altitudes. A-J. spring cases. K-N. fall cases. See text for
additional details.
196
SIDNEY A. GAUTHREAUX, JR.
FIG. 9. ASR-7 radar displays of echoes from migrating songbirds at night (range-rings 2 nautical miles or
3.7 km apart). Upper—7 May 1973, Greenville, SC; 20:18 EST. Showing most frequent altitudinal distribution. Lower—5 May 1973, Greenville, SC; 22:10 EST. Showing migration at higher altitude than usual.
tudinal measurements from the radar display undoubtedly contributed to the poorer
correlation between the altitudes of densest migration and the altitudes of most
favorable winds. When one examines the
flight directions of these higher altitude
movements in relation to the winds at the
same altitudes, the two are closely associated (Fig. 12). In this figure spring flights
above 500 m are plotted in the right diagram; the corresponding winds on these
occasions are plotted on the left. The mean
MIGRATING BIRD FLIGHT BEHAVIOR
oo
10
i i i i i i i i i i i i i i i i I i i i i
50
20
30
40
FREQUENCY
FIG. 10. Altitudinal distribution of nocturnal
migrants on 184 nights in spring and 13 nights in fall
as measured with ASR-7 radar.
wind direction is from 215° (r = 0.73, s =
45°) and blowing toward 35°; the mean
direction of migrations is toward 31° (r =
0.85, s = 32°).
Additional studies
Several spring moon-watching studies of
the altitudinal distribution of migrants aloft
in relation to winds aloft have been summarized by Dolnik and Bolshakov (1985).
The studies examined the height and
direction of spring nocturnal migration
over the 2,200 km wide zone from the Caspian Sea to the eastern part of the Tien
Sham mountains, including the Karakum
and the Kizilkum deserts and the Copetdag, Tien Sham, and Pamirs areas. Radiosondes were taken at five sites throughout
the study area. The ranges of altitudes with
the greatest density of migration were 300800 m above ground level (agl) over the
foothill areas, 1,100-1,200 m agl over
mountain areas, and 300-950 m agl over
desert areas. Passerine birds represented
73.4-88.8% of the total nocturnal migration recorded. Dolnik and Bolshakov suggested that the relatively low height of
migration over the foothills and mountain
areas was due to a) low-altitude mass migration of birds along mountain ridges and in
reverse directions from the lowland, b)
197
0
1
2
3
ALTITUDE OF MOST FAVORABLE WINDS ALOFT (KM)
FIG. 11. Correlation between the altitude of most
favorable wind aloft and the median altitude of nocturnal migration when flights were above an altitude
of 500 m agl.
descending of birds during the night, c)
increased descending with unfavorable
wind conditions over the mountains, and
d) inclusion of birds just beginning migration. The major directions of movement
were toward the NW-N from wintering
areas in southern Asia and toward the NE
from Africa and the southwestern part of
Asia. Movements toward the ESE-E
occurred everywhere except the eastern
shore of the Caspian Sea and resulted from
three processes: a) deviation from NE
directions by strong NW winds, b) compensation movements after previous lateral drift, and c) reactions of some high
flying birds to high ridges.
FIG. 12. Flight directions and wind directions when
flights were above 500 m agl. A. Wind directions at
median altitudes of nocturnal migration. B. Directions of migrations above 500 m agl.
198
SIDNEY A. GAUTHREAUX, JR.
Similar studies of the flight strategies of
migrating birds in autumn over the central
Asian mountains have been summarized by
Dolnik (1985). Nocturnal migration was
observed by moon-watching from 19781985 at 10 sites located in the mountain
systems of central Asia, and radiosonde
measurements of the winds aloft were taken
at four localities. The flights over the
mountains were higher than in the spring
and higher than flights over the lowland
deserts in the autumn. The general direction of migration was toward the SSW
(215°). The dominant high-altitude wind
from the west changed the direction of passerine migrants 30°-40° to the S and SSE.
Many birds crossed the mountains at
heights 1-3 km above the average altitude
of the mountain range. In addition to wind
influences, topography strongly affected
the local directions of migratory flight, the
altitude of the flight, and the concentration of migrants near mountain ranges.
The influence of local topography and
winds aloft on the nocturnal migration of
birds at the border of the Alps has been
examined by Liechti and Bruderer (1986).
More than 10,000 individual tracks of nocturnal migrants were recorded with
"Superfledermaus" radar in the autumns
of 1980 and 1982. When winds were from
the side or opposed to the principal direction of migration (basically toward the
southwest in autumn) the low flying
migrants flew parallel to the slopes of the
mountains, while migrants just above the
ridges followed the ridges in the sense of
a leading line, and migrants above the
ridges, particularly when above clouds,
showed drift or down-wind flights. In light
or northeasterly winds the difference in the
direction of movement of low and high flying migrants was reduced. Liechti and Bruderer suggest that the flight strategy used
by migrants to overcome an obstacle consists of optimizing between minimum climb
and minimum deviation in direction. With
increasing flight level, the tendency for
climbing and shifting direction in response
to topographic obstacles decreases, and the
migrants are less likely (or capable) to fly
against the wind or to compensate for drift
at high altitudes, even if the wind speed is
the same at high and low levels.
The altitudinal aspects of daytime broad
front migration over East Netherlands had
been examined by Buurma et al. (1986). In
their x-band tracking radar and visual study
of autumn daytime migration in 1984, they
found that for many of the daytime
migrants the proportion of birds over
thrush (Turdus) size increases with altitude,
and larger birds are less hindered by headwinds. Only 17% of 359 radar records
occurred above 200 m. Low level migration is extra low over bare country while
wind effects caused by roughness of the
landscape enable birds to fly higher at certain locations.
Additional details of atmospheric structure, particularly boundary layer thickness, turbulence, and convection, and how
these influence the altitude and timing of
bird migration during the day and at night
can be found in Kerlinger and Moore
(1989). They analyzed microscale atmospheric data on vertical and horizontal
winds, temperature, and relative humidity
fluctuations for two locations in North
America (Albany, NY and Lake Charles,
LA) and generated predictions as to the
optimal time of day and altitudinal strata
for migration. They concluded that four
characteristics of the nocturnal atmosphere make migration more favorable for
powered migrants than characteristics of
the atmosphere during the daylight: 1)
cooler night air temperatures, 2) slower
horizontal winds after dark, 3) less variable
winds after dark because of reduced thermal generated turbulence, and 4) the
absence of vertical wind currents caused
by thermal convection. Kerlinger and
Moore provide a literature review of air
speed, altitude, and diel timing of passerine, waterfowl, shorebird, and soaring bird
migration that supports their hypothesis
showing that birds that use powered flight
migrate at night or in the morning before
10:00 hr, and soaring migrants fly when
thermal conditions are best near midday.
They find that large, fast, powered flyers
(waterfowl and shorebirds) are more likely
to fly at higher altitudes where winds are
MIGRATING BIRD FLIGHT BEHAVIOR
stronger and during the day especially during late autumn and early spring when temperatures are cooler and less turbulence is
present.
There can be little doubt that the variance in the altitudes of migratory flights
of songbirds during the day and at night
is clearly related to the dynamics of the
atmosphere. The deviations in altitude of
flight are generally associated with winds
at those altitudes that are favorable to
movement in the appropriate seasonal
directions (either largely following winds
or minimal head winds) suggesting that
birds alter their altitudes of flight to take
advantage of favorable winds so as to minimize the cost of transport. Although not
much is known about how birds are able
to find the flight level where flying conditions are most favorable, the question of
migrants recognizing optimum flight level
has been recently addressed by Elkins
(1988). He proposes that high-altitude
migrants can recognize favorable wind
flows by 1) sensing vertical wind shear
(VWS), that is the change in wind velocity
and direction with height, or 2) recognizing the distinctive cloud formations associated with VWS and jet streams, or 3)
detecting infrasound made by jet streams.
The flight behavior of migrating birds
when encountering changing wind fields
associated with mesoscale atmospheric
phenomena such as frontal passages and
buoyancy waves during the spring and fall
shed some light on how migrants might
find more favorable flying conditions.
Flight behavior, fronts, and
atmospheric waves
Since the beginning of autumn 1964, I
have examined the distribution of nocturnal migrants aloft in terms of a change in
altitude, flight direction, or both, during
the passage of cold fronts or when wind
directions at the surface and aloft deviated
from the normal seasonal direction of
migration. Such deviations characteristically occur in spring just after the passage
of a cold front with air flow at lower altitudes from the northwest, north, or northeast and in fall ahead of a cold front with
199
air flow from the southeast, south, or
southwest. Although wind changes aloft are
most frequently associated with frontal
passages and the movement of atmospheric
pressure systems, atmospheric thermal
exchange at night in relation to topography may also cause wind changes aloft (e.g.,
nighttime sea breezes and buoyancy waves).
I have gathered data during 32 cold
fronts (10 strong and 22 weak) and five
wind changes aloft without a front in the
spring; and nine cold fronts (one weak and
eight strong) and 10 cases when winds aloft
changed in the absence of frontal systems.
The flight tactics of migrant birds aloft
during the passage of a front differs
depending on the strength and depth of
the mass of cold air and on the season. In
vertical cross-section the slopes of cold
fronts (the boundary between the denser,
colder air and the lighter, warmer air) range
from as little as 1:250 km (0.23°) to as steep
as 1:50 km (1.14°) with an average of 1:150
km (0.37°), and the frontal zone is usually
about 50 to 100 km in extent (Miller and
Thompson, 1970). In spring when a strong
cold front without precipitation passed
through the area of radar surveillance and
radar studies of bird migration were
underway, the responses of the migrants
aloft were clearly visible on the radar
screens (plan position indicator—PPI and
range height indicator—RHI). Prior to
frontal passage the winds blow from a
southwesterly direction and are favorable
for the northeastward movement of birds.
As the leading edge of the colder air moves
over the radar station and the winds shift
and come from a northwesterly direction,
the pattern of echoes from passerine
migrants aloft shows two changes. These
changes are illustrated with drawings in
Figure 13A and B. The migrants appear
to gain altitude at the leading edge of the
front, and their direction of movement
shifts toward a more easterly or east-southeasterly direction (Fig. 13A). Behind the
leading edge of the front only a few larger
echoes from birds remain, and these continue to move toward the northeast (Fig.
13B). The latter echoes have been identified on only one occasion as aflockof ducks,
200
SIDNEY A. GAUTHREAUX, JR.
20
10
0
10
20
FIG. 13. Composite drawings based on WSR-57 radar displays and photographs showing the responses of
migrating birds during cold front passage at night. The dots represent echoes from birds. The range marks
are 5 nautical miles (9.3 km) apart and the antenna elevation is 3°. A-B. Spring with strong cold front. CD. Fall strong cold front.
so that most of the larger echoes are produced by stronger flying waterfowl (Anseriformes) and shorebirds (Charadriiformes). Once the frontal edge has moved
approximately 20 nautical miles beyond the
radar station, a line of fine echoes at higher
altitudes (birds climbing at the edge of the
front?) is still visible (Fig. 13B), but virtu-
ally all songbird migration behind the front
has ceased.
In the fall cold fronts usually stimulate
dense migrations of passerines when the
fronts pass through the area of radar surveillance during the first half of the night.
Prior to frontal passage relatively few
migrants are displayed on the radar and
MIGRATING BIRD FLIGHT BEHAVIOR
most of these are moving toward the northeast with winds from the southwest. As the
front moves over the radar station and
winds shift to blow from the northwest, the
density of songbird migration increases
dramatically. On at least five occasions it
was possible to monitor the wind shift line
on the PPI because of the sharp line of
demarcation in the pattern of bird echoes
(Fig. 13C). One hour after the passage of
the front, the radar screen (PPI) shows an
increase in the number of songbird
migrants (Fig. 13D).
The flight responses of migrant birds
aloft appear to be very different in the
spring when weak, shallow cold fronts without precipitation move through the area
of radar surveillance. Prior to the passage
of the front a good migration is usually
underway with following winds. As the
front passes over the radar station and the
lower winds shift from a southwesterly
direction to a northwesterly one, the altitude of the passerine migrants increases
(Fig. 14A and B). The migrants do not
change their direction of flight but continue moving toward the northeast with
following winds above the wedge of colder
air moving toward the southeast. Thus the
migrants by slowly gaining altitude continue to have favorable flying conditions.
The resultant display of echoes on the PPI
shows a distinct "doughnut" configuration, a display that is characteristic when
migrants are flying over a stratum of air
largely empty of migrants (Fig. 14C and D).
A radiotelemetry study of migratory
thrushes (Catharus spp.) by Cochran and
Kjos (1985) confirms many of the preceding findings from surveillance radar and
offers some of the best information available on the flight tactics of migrating birds
in changing wind fields. They found that
individual thrushes do not maintain a constant track direction unless the wind they
are in is constant, and the thrushes maintain a constant heading during a night's
flight and from night-to-night to an accuracy equal to or better than ±3°. Occasionally thrushes changed their headings
in response to lateral wind but at a range
of change less than 3° per hr, and if they
alter their airspeeds to reduce drift they
201
do so by less than 2 m/sec. Thrushes minimize drift by lateral winds by flying at altitudes where winds are favorable provided
progress speed does not fall below about 6
m/sec. When the birds must choose
between lateral drift up to 60° and progress
speed below 6 m/sec, the birds accept lateral drift. If winds at all altitudes above
about 75 m have unfavorable head and side
components so that progress speed cannot
be maintained above 2 or 3 m/sec, the
birds lands.
After takeoff, Cochran and Kjos found
that thrushes ascend until they find suitable winds and if favorable winds are not
found below 2 or 3 km, the birds descend
and either land or accept a compromise at
a lower altitude, but never below about 75
m. As winds change during a night's flight,
the thrushes adjust their altitude accordingly, usually by descending and rarely by
ascending. During the study the birds never
ascended to an altitude reached during the
initial ascent at the beginning of a night.
The ascent rate of the thrushes at the
beginning of the night is usually in the
range of 0.5-1.0 m/sec, a rate that is less
than their climbing capability, and Cochran and Kjos believe that this rate may
reflect the time needed for assessment of
winds aloft. They suggest that long migrations made up of many shorter single night's
flights are subject only to net drift according to the prevailing wind components of
the various geographic areas traversed, and
they found that the net drift was remarkably small for short path distances of 400—
700 km even though the cases selected for
presentation showed drift more than most
of the tracks they recorded. Cochran and
Kjos conclude that constant-heading
behavior with wind drift mitigated by altitude changes to select a not-too-unfavorable wind, represents a long-distance
migratory strategy that, in view of the
probability that net drift is negligible, is
more conservative of energy than strategies involving complete or partial compensation for wind.
The use of a constant compass heading
during long migratory flights has been
computer simulated for the autumnal
migration of birds over the western North
202
SIDNEY A. GAUTHREAUX, JR.
20
10
0
10
20
20
10
0
10
20
FIG. 14. Composite drawings based on WSR-57 radar displays and photographs showing the responses of
migrating birds during the passage of weak, shallow cold fronts in spring. A-D show the formation of the
"doughnut" display of bird echoes when the birds are flying higher than usual. See Figure 13 for additional
information.
Atlantic (Stoddard et al., 1983). By maintaining a constant compass heading and
allowing natural deflection by the wind to
guide their course, many shorebirds and
passerines could depart from the North
American coast and successfully reach their
probable destination on the South Amer-
ican coast. Airspeeds, headings, altitude of
flight, and point of departure have major
effects on the success of the simulated flight.
SUMMARY AND CONCLUSIONS
In two previous papers I examined the
underlying climatic and ecological factors
203
MIGRATING BIRD FLIGHT BEHAVIOR
TABLE 2.
Category
Microscale
Mesoscale
Synoptic
Global
Scales of atmospheric motion.*
Phenomenon
Turbulence
Convection
Gamma (thunderstorms)
Beta (squall lines)
Alpha (fronts and hurricanes)
Baroclinic waves (cyclones)
Seasonal circulations (monsoon)
Lagrangian time scale
1 sec—2 min
2 min-20 min
10 min-2 hr
2 hr-1 day
1 day—7 days
2 days-1 month
3 months-1 year
Spatial scale
1 mm-1 m
50 m-2 km
5 km-20 km
20 km-200 km
200 km-2,000 km
2,000 km-5,000 km
>5,000 km
* After Orlanski, 1975.
that have interacted and influenced the
evolution of migratory patterns in birds
(Gauthreaux, 1980a, 1982). In this paper
I have emphasized the more direct influence of atmospheric structure and motion
on the flight behavior of migrating birds.
The climatic and meteorological factors
responsible for changing the suitability of
a habitat through time may be in large
measure the same factors that shape the
spatial and temporal characteristics of
movements from and to that habitat. By
concentrating on atmospheric transport as
the geophysical variable of interest, a number of other climatic variables that influence the temporal and spatial distribution
of migrants (e.g., temperature, barometric
pressure, humidity) can be integrated. I
examined the influence of atmospheric
structure on migration by emphasizing
radar studies of the distribution of migrants
aloft above the radar station during the day
and at night and during the passage of
frontal systems. Consequently, the
responses of migrating birds to changing
wind fields is examined at local and regional
spatial scales, and one can extrapolate to
the continental spatial scale (see Gauthreaux, 1980c). The findings suggest that
migrating birds aloft are sensitive to
changes in atmospheric structure and
motion at greatly differing scales and that
flight strategies have evolved in response
to a complex array of atmospheric dynamics operating hierarchically in space and
time.
ability for survival or breeding to areas of
increasing suitability. The scales of climatic impacts range from 1 mm-1 m to
greater than 5,000 km and at Lagrangian
time scales ranging from 1 sec to months
(Table 2). The atmospheric phenomena
range from microscale turbulence and convection thermals and plumes (tremendously important to soaring and drift
migrants), through mesoscale events like
thunderstorms, squall lines, and sea
breezes, to larger scale synoptic and seasonal patterns of baroclinic waves or
cyclones. The spatial and temporal scales
of climatic phenomena as they relate to
various ecological and social processes have
been discussed and diagrammed in Clark
(1985). The duration of the climatic events
is also given. Thus the major circulation
patterns of the atmosphere not only influence the location and duration of habitats
that are important to migrants, but the circulation patterns clearly influence directly
the flight strategies and migratory pathways of atmospheric migrants. An analysis
of global atmospheric circulation below an
altitude of 6 km could reveal the atmospheric factors that have shaped and continue to shape the migratory strategies of
long-distance migrants. Clearly more
detailed radar and radiotelemetry studies
of migrants aloft are needed so that the
patterns of migrant-atmospheric interactions can be thoroughly documented.
Atmospheric processes operating simultaneously at vastly different spatial and
temporal scales have greatly influenced the
flight tactics of migrants aloft because of
the need to minimize the energetic cost of
moving from habitats of declining suit-
Able, K. P. 1977. The flight behavior of individual
passerine nocturnal migrants: A tracking radar
study. Animal Behaviour 25:924-935.
Able, K. P. and S. A. Gauthreaux, Jr. 1975. Quantification of nocturnal passerine migration with
a portable ceilometer. Condor 77:92-96.
Alerstam, T. 1985. Strategies of migratory flight,
REFERENCES
204
SIDNEY A. GAUTHREAUX, JR.
illustrated by arctic and common terns, Sterna Gauthreaux, S. A., Jr. 1980c. The influence of global
climatological factors on the evolution of bird
paradisaea and Sterna hirundo. Contributions in
migratory pathways. Proc. XVIIth Intern. OrniMarine Science (Supplement) 27:580-603.
thol. Congress. West Berlin, pp. 517-525. Verlag
Bruderer, B. and P. Steidinger. 1972. Methods of
der Deutschen Ornithologen-Gesellschaft, Berquantitative and qualitative analysis of bird
lin.
migration with tracking radar. In S. R. Caller, K.
Schmidt-Koenig, G. J. Jacobs, and R. E. Belleville Gauthreaux, S. A., Jr. 1982. The ecology and evo(eds.), Animal orientation and navigation, pp. 151lution of avian migration systems. In D. S. Farner
167. National Aeronautics and Space Admin.
and J. R. King (eds.), Avian biology, Vol. VI, pp.
93-168. Academic Press, New York.
Special Publ. 262.
Buurma, L. S., R. Lensink, and L. G. Linnartz. 1986. Gauthreaux, S. A., Jr. 1985. Radar, electro-optical,
and visual methods of studying bird flight near
Hoogte van breedfronttrek overdag boven
transmission lines. Electric Power Research InstiTwente: een vergelijking van radar en visuele
tute, Ecological Studies Program, Report EAwaarnemingen in oktober 1984. Limosa 59:1694120. 76 pp.
182.
Clark, W.C. 1985. Scales of climatic impacts. Climate Kerlinger, P. and F. R. Moore. 1989. Atmospheric
structure and avian migration. Current OrniChange 7:5-27.
thology, 6:109-142. Plenum Press, New York.
Cochran, W. W. 1972. Long-distance tracking of
birds. In S. R. Galler, K. Schmidt-Koenig, G. J. Larkin, R. P. and D. Thompson. 1980. Flight speeds
of birds observed with radar: Evidence for two
Jacobs, and R. E. Belleville (eds.), Animal orienphases of migratory flight. Behavioral Ecology
tation and navigation, pp. 39—59. National Aeroand Sociobiology 7:301-317.
nautics and Space Admin. Special Publ. 262.
Cochran, W. W. and C. G. Kjos. 1985. Wind drift Liechti, F. and B. Bruderer. 1986. Einfluss der lokalen Topographie auf nachtlich ziehende Vogel
and migration of thrushes: A telemetry study.
nach Radarstudieen am Alpenrand. Der OrniIllinois Natural History Survey Bull. 33:297-330.
thologische Beobachter 83:35-66.
Cochran, W. W., G. G. Montgomery, and R. R. Graber. 1967. Migratory flights of hylocichla Lowery, G. H.,Jr. 1951. A quantitative study of the
nocturnal migration of birds. Univ. Kansas Publ.
thrushes. Living Bird 6:213-225.
Mus. Nat. Hist. 3:361-472.
Dolnik, V. R. (ed.) 1985. Nocturnal bird migrations
Miller, A. and J. C. Thompson. 1970. Elements of
over arid and mountain regions of Middle Asia
meteorology. Charles E. Merrill Publ. Co., Columand Kazakhstan. Proc. Zool. Inst., Leningrad
bus, Ohio.
(USSR Acad. Sci.) 138:1-150. [In Russian, English
Nisbet, I. C. T. 1959. Calculation of flight directions
summary]
of birds observed crossing the face of the moon.
Dolnik, V. R. and K. V. Bolshakov. 1985. PrelimiThe Wilson Bulletin 71:237-243.
nary results of vernal nocturnal bird passage study
over arid and mountain areas of central Asia: Orlanski, I. 1975. A rational subdivision of scales for
atmospheric processes. Bull. Amer. Meteor. Soc.
Latitudinal crossing. In V. R. Dolnik (ed.), Spring
nocturnal bird passage over arid and mountain areas
56:529-530.
ofAsia Middle and Kazakhstan, pp. 260-294. USSR Richter,J. H., D. R.Jensen, V. R. Noonkester,J. B.
Acad. Sci., Zool. Inst., Leningrad. [In Russian,
Kreasky, M. W. Stimmann et al. 1973. Remote
English summaries]
radar sensing: Atmospheric structure and insects.
Science 180:1176-1178.
Drake, V. A. and R. A. Farrow. 1988. The influence
Schaefer, G. W. 1976. Radar observations of insect
of atmospheric structure and motions on insect
flight. Symp. Royal Entomol. Soc. London 7:157migration. Ann. Rev. Entomol. 33:183-210.
197.
Eastwood, E. 1967. Radar ornithology. Methuen, LonSchaefer, G. W. 1979. An airborne radar technique
don.
for the investigation of control of migrating pest
Elkins, N. 1988. Can high-altitude migrants recoginsects. Philos. Trans. Royal Soc. London Ser. B
nize optimum flight levels? Ibis 130:562-563.
287:459-465.
Gauthreaux, S. A., Jr. 1969. A portable ceilometer
Sparks, A. N., J. K. Westbrook, W. W. Wolf, W. S.
technique for studying low-level nocturnal
Pair, and J. R. Raulston. 1985. Atmospheric
migration. Bird-Banding 40:309-320.
transport of biotic agents on a local scale. In D.
Gauthreaux, S. A., Jr. 1980a. The influence of longR. MacKenzie, C. S. Barfield, G. C. Kennedy, R.
term and short-term climatic changes on the disD. Berger, and D. J. Taranto (eds.), The movement
persal and migration of organisms. In S. A.
and dispersal ofagriculturally important biotic agents,
Gauthreaux, Jr. (ed.). Animal migration, orientapp. 203-217. Claitor, Baton Rouge, Louisiana.
tion, and navigation, pp. 103-174. Academic Press,
New York.
Stoddard, P. K., J. E. Marsden, and T. C. Williams.
1983. Computer simulation of autumnal bird
Gauthreaux, S. A., Jr. 19804. Direct visual and radar
migration over the western North Atlantic. Animethods for the detection, quantification, and
mal Behaviour 31:173-180.
prediction of bird migration. Special Publication
No. 2, Department of Zoology, Clemson University, Clemson, South Carolina 29631. 67 pp.