Nocturnal autumn bird migration at Falsterbo, South Sweden

Abstract

We investigated the patterns of nocturnal bird migration in autumn 1998 at a coastal site on the Falsterbo peninsula in south-western Sweden, by means of a passive infrared device. In total 17 411 flight paths, including track direction and altitude, of migrating birds were recorded for 68 nights from August to October. Mean migratory traffic rate per night varied between 6 and 6618 birds km1 h1, with an average of 1319 birds km1 h1. Migration at Falsterbo showed a similar seasonal pattern to that reported for central Europe, with pronounced peaks of migration and intermittent periods with relatively low migratory intensities. Weather factors explained two thirds of the variance in the intensity of bird migration. During nights with intense migration, associated with weak winds, the mean track direction was close to that in central western Europe (225°). Birds usually maintained a constant heading independent of wind directions and, in consequence, were drifted by the wind. The mean orientation clearly differed from that of the nearest coastline, suggesting that the birds did not use the topography below to compensate for wind drift.

Full text

JOURNAL OF AVIAN BIOLOGY 32: 239–248. Copenhagen 2001
Nocturnal autumn bird migration at Falsterbo, South Sweden
Susanna Zehnder, Susanne A
,
kesson, Felix Liechti and Bruno Bruderer
Zehnder, S., A
,
kesson, S., Liechti, F. and Bruderer, B. 2001. Nocturnal autumn bird
migration at Falsterbo, South Sweden. – J. Avian Biol. 32: 239–248.
We investigated the patterns of nocturnal bird migration in autumn 1998 at a coastal
site on the Falsterbo peninsula in south-western Sweden, by means of a passive
infrared device. In total 17411 flight paths, including track direction and altitude, of
migrating birds were recorded for 68 nights from August to October. Mean migra-
tory traffic rate per night varied between 6 and 6618 birds km
−1
h
−1
, with an
average of 1319 birds km
−1
h
−1
. Migration at Falsterbo showed a similar seasonal
pattern to that reported for central Europe, with pronounced peaks of migration and
intermittent periods with relatively low migratory intensities. Weather factors ex-
plained two thirds of the variance in the intensity of bird migration. During nights
with intense migration, associated with weak winds, the mean track direction was
close to that in central western Europe (225°). Birds usually maintained a constant
heading independent of wind directions and, in consequence, were drifted by the
wind. The mean orientation clearly differed from that of the nearest coastline,
suggesting that the birds did not use the topography below to compensate for wind
drift.
S.Zehnder
(
correspondence
)
,Swiss Ornithological Institute,CH-
6204
Sempach,
Switzerland and Institute of Zoology,Uni6ersity of Zu¨rich,Winterthurerstr.
190
,
CH-
8057
Zu¨rich,Switzerland.E-mail
:
Susanna.Zehnder@Vogelwarte.ch.S.A
,
kesson,
Department of Animal Ecology,Lund Uni6ersity,Ecology Building,SE-
22362
Lund,
Sweden.F.Liechti and B.Bruderer,Swiss Ornithological Institute,CH-
6204
Sempach,
Switzerland.
The migratory behaviour of birds is controlled by en-
dogenous factors and influenced by various environ-
mental factors. Important exogenous factors are
weather and topography (Alerstam 1976a). At the mid-
latitudes in Europe the seasonal pattern of bird migra-
tion is strongly influenced by the itinerant low-pressure
systems. Heavy autumn migration occurs mainly to the
east of high-pressure areas, often immediately after a
depression has passed (Richardson 1978, 1990, Aler-
stam 1990, Bruderer 1997). Alerstam (1990) stated that
once wind and precipitation conditions are taken into
account to explain the intensity of migration, other
factors do not play any major role. It has been widely
established that wind has a paramount effect on migra-
tory patterns, flight directions and altitudes (Alerstam
1979, Richardson 1991). Weak tail winds lead to large
numbers of migrants aloft. Wind drift and compensa-
tion have mainly been investigated in relation to topog-
raphy (e.g. Koch 1934, Williams et al. 1977, Bingman et
al. 1982, A
,
kesson 1993a, Liechti 1993, Liechti et al.
1996, Bruderer 1997, Bruderer and Liechti 1998a, b).
Birds appear to follow topographical structures to com-
pensate for wind drift. Mountain ridges, valleys, rivers
or coastlines are often used by diurnal migrants as
leading lines, but have as well been shown to be used by
nocturnal migrants for wind compensation (Bingman et
al. 1982, A
,
kesson 1993a, Bruderer and Liechti 1998b).
The effect of a leading line increases the more it is
orientated in the general migratory direction and the
lower birds are flying (Bruderer et al. 1989).
Falsterbo, at the south-western tip of Sweden, is the
most famous site for observing the passage of diurnal
autumn migration in Scandinavia (Alerstam and Ulf-
strand 1972, 1975, Alerstam 1978). Each year thou-
sands of passerine migrants are ringed at Falsterbo Bird
Observatory (Karlsson 1993). The orientation of noc-
turnal passerine migrants captured at this site has been
investigated in orientation cage experiments (e.g. Sand-
berg et al. 1988, A
,
kesson 1993b, 1994, A
,
kesson and
Ba¨ckman 1999) and by ringing recovery analysis (e.g.
Sandberg et al. 1988, A
,
kesson et al. 1996b). However,
little is known about the nocturnal passage and those
© JOURNAL OF AVIAN BIOLOGY
JOURNAL OF AVIAN BIOLOGY 32:3 (2001) 239

Fig. 1. Map of southern Sweden. The star on the inset shows
the location of the observation site on the Falsterbo peninsula
(55°23%N, 12°50%E).
migratory intensity, flight directions and altitudes are
analysed in relation to wind and topography. We inves-
tigated whether birds migrating at high altitudes, like
low altitude migrants (cf. A
,
kesson 1993a), flew along
the coastline to compensate for wind drift.
Study site and methods
Nocturnal migration was monitored by using an in-
frared device at the southern tip of the Falsterbo penin-
sula (55°23%N, 12°50%E; 5 m above sea level, ASL) in
south-westernmost Sweden (Fig. 1). The observation
point was located at the Falsterbo Lighthouse, which is
surrounded by a small stand of bushes and trees in
otherwise open country.
Observations were carried out between 7 August and
28 October 1998, hence they covered the main autumn
passage (Karlsson 1993). If not prevented by precipita-
tion, the recordings started one hour before sunset and
lasted until one hour after sunrise. The overall noctur-
nal pattern showed two minima of migration intensity,
the first approximately at sunset and the second 80 min
before sunrise. In order to exclude the potentially con-
founding influence of diurnal movements, which in-
creased considerably already before sunrise, the
analyses were restricted to the time interval between the
end of civil twilight at dusk and the beginning of civil
twilight at dawn (Fig. 2). According to this definition,
nights lasted from 21:53 to 4:34 h (6.7 h) in early
August, and from 18:24 to 7:20 h (13.0 h) at the end of
October (UTC+2 h).
The thermal imaging equipment consisted of a pas-
sive infrared device (Long-Range-Infrared System
LORIS, IRTV-445L, Inframetrics, Massachusetts,
USA), a video recorder, a video peak store (Video Peak
Store 493, Colorado Video Inc.), and a monitor. The
camera with an opening angle of 1.45°was placed in a
vertical upward position on the ground and was ad-
justed towards north with the aid of a compass. The
camera was focused periodically on an object a few
kilometres away in order to ensure a well-focused im-
age from 300 m to infinity. Thermal radiation emitted
studies addressing it have been done for short periods
only, using either surveillance radar (Alerstam 1972,
1976b) or observations of birds flying through a vertical
light beam (ceilometer study, A
,
kesson 1993a). A
,
kesson
(1993a) showed that nocturnal passerine migrants flying
at low altitude follow the coastlines of the outermost
parts of the Falsterbo peninsula and use the coastline to
compensate for wind drift.
In this paper we present an almost complete seasonal
survey of the course of nocturnal autumn migration at
Falsterbo based on absolute numbers of migrating
birds. We recorded the temporal pattern of nocturnal
migration at a coastal site on the Falsterbo peninsula
with a passive infrared device to study its variation with
weather conditions. Seasonal and nightly patterns of
Fig. 2. Nocturnal pattern of mean
MTR relative to sunset and
sunrise, based on pooled data from
the entire season. The arrows
indicate the end of civil twilight at
dusk and the beginning of civil
twilight at dawn, respectively.
240 JOURNAL OF AVIAN BIOLOGY 32:3 (2001)

Fig. 3. (A) Monitor showing the
picture of a night with cloud
passage. Clouds and birds appear
dark in contrast to the cold sky. A
single bird of size 4 is visible near
ten o’clock. (B) Monitor showing a
peak store picture (similar to a long
term exposure). Each line of dots
corresponds to a single bird. The
flight direction of the bird is
determined according to the closest
numbers of entrance and exit of the
clock face. The bird’s altitude is
estimated according to the scale
marked in the lower right hand
corner.
by any kind of material can be detected if contrasting
by at least 0.06°C against the background. Birds are
normally seen as unstructured silhouettes. A bird of the
size of a small passerine can be detected from very low
altitudes up to 3000 m above ground level (AGL) as
has been shown by direct comparisons with tracking
radar (Bruderer and Liechti 1994, Liechti et al. 1995).
Birds flying below 300 m AGL are increasingly out of
focus, but clearly visible down to the lowest observed
flight levels at roughly 25 m AGL (direct comparison
with the height of the lighthouse).
Clouds seriously impair applicability of the technique
because they contrast strongly with the clear sky and
birds cannot be detected through them (Fig. 3A).
Therefore, a cloudless sky meets the requirements for
optimal observations. Because the infrared picture is
built on heat differences, completely overcast and clear
skies cannot be distinguished on the screen. The base of
the clouds represents the upper limit for bird detection.
All observations were recorded on video tapes and later
analysed, for cloudless and completely overcast skies by
means of video peak store. This device produces a
picture on the screen similar to a long-term exposure. A
bird’sflight path appears as a line of dots, which is
easier to observe than a moving silhouette crossing the
screen (Fig. 3B).
Sometimes bird-shaped silhouettes or indications of
flapping phases can be seen, but the identification of
species remains impossible. Therefore, it cannot be
excluded that a very small fraction of the observations
is due to hunting or migrating bats (e.g. Ahle´n 1997).
Insects are much more numerous than bats but radiate
less heat. Even large insects produce roughly 10 times
less heat than small birds. Heat radiation decreases by
distance squared. This suggests that even for large
insects detectability would decline rapidly beyond 30 m.
Comparisons with parallel radar measurements in
Switzerland gave no indication that insects detected by
radar (min. distance 100 m) could be detected by the
infrared device (unpubl. data). Due to the narrow beam
at short distances the passage through the field of view
is very fast, at least for large insects. Medium-sized
insects would only be visible up to a few metres; they
would, therefore, be much out of focus and provide
relatively large bubble-like targets. Such targets were
excluded from the counts. Small and weak fast-moving
targets were also considered to be insects and were
excluded. The quality of the picture deteriorates at
sunset and sunrise due to the sun’s infrared radiation.
Hence, the quality of thermal imaging is severely re-
duced under daylight conditions and cannot be com-
pared quantitatively with nocturnal observations.
The higher a bird flies above the camera the smaller
the silhouette appears on the screen and the longer it
takes to cross the field of view. Silhouettes were
classified into seven size classes. The largest (class 7)
was often out of focus, but was much darker than the
bubble-like targets classified as insects. Classes 6 to 1
were compared with a reference scale (points of differ-
ent size) mounted on the screen (Fig. 3). The estimates
of flight altitude for each size class, according to radar
measurements, are given in Table 2. The vast majority
of nocturnal migrants are passerines of the size between
a warbler and a thrush (Bloch et al. 1981). The size
differences between the real silhouettes vary, therefore,
within narrow limits. In contrast, the heat transmitted
towards the infrared device decreases at least with the
square of the distance in clear air. Thus, the usual size
differences between various passerines are negligible
compared to the influence of distance.
The flight directions, i.e. track directions, were deter-
mined according to the clock face method suggested by
Lowery and Newman (1955) for the moon-watching
technique (12 o’clock=N;6o’clock =S; clockwise). A
circle was marked on the screen (see Fig. 2). For each
bird the closest full hours of entrance and exit were
registered; thus, 24 different flight directions were possi-
ble. Only birds flying through the circle were analysed.
The circle on the screen corresponded to 1.4°opening
angle of the field of view. The conical shape of the
controlled space leads to an altitude-dependent detec-
tion probability of birds aloft. Therefore, numbers of
observed birds were transformed into migration traffic
rates (MTR) (Lowery 1951, adjusted to the metric
JOURNAL OF AVIAN BIOLOGY 32:3 (2001) 241

system). The MTR gives the number of birds crossing a
line of one kilometre perpendicular to the flight direc-
tion during one hour (birds km
−1
h
−1
), calculated
according to the following approximation:
MTR=
%
7
i=1
x
i
·(d
i
·(sin f))
−1
Dt(1)
where i=height classes 1 to 7, x
i
=number of birds in
height class i, d
i
=flight altitude (km), f=opening
angle of the camera and Dt=observation interval (h).
For each night the mean MTR was calculated. For
specific analyses, nights were classified into three inten-
sity classes. The average night included 1.4% of the
seasonal passage. Therefore, we defined intense migra-
tion nights as those including \2% of total passage of
the season, intermediate nights as those with 0.5–2%
and low migration nights as those with B0.5%. The
sum of MTR of the intense migratory nights (N=16)
accounted for 73% of the total passage, intermediate
nights (N=20) for 21% and low migratory nights
(N=32) for 6%.
Weather data for Falsterbo were provided by the
Swedish Meteorological and Hydrological Institute
(surface wind direction and speed, temperature, dew
point temperature, altitude of cloud base, degree of
cloud cover, air pressure, visibility; 3-hourly measure-
ments). For all analyses mean values based on the 21,
00 and 03 UTC measurements were used. The 24-h
changes of temperature and air pressure were calculated
as the difference between the measurements at 00 UTC
of two consecutive nights. Surface winds were used to
calculate the headings of birds at 0–1000 m ASL. This
extrapolation is possible due to the plain topography.
For the headings of birds at 1001 m–3000 m ASL the
average wind direction and speed was calculated based
on the 23 UTC radio sonde from Kastrup Airport,
Denmark (25 km NNW of Falsterbo).
The track of a bird over the ground is the vector sum
of its heading (direction and speed with respect to the
air) and the wind (direction and speed). Hence, mean
heading directions were calculated as follows:
Heading=a−arcsinv
w
·sin(a−v)
v
a
(2)
where a=mean track direction, v=wind direction,
v
w
=wind speed (m s
−1
) and v
a
=average air speed of
the bird (m s
−1
). For these calculations the bird’s air
speed was estimated at 12 m s
−1
(Liechti and Bruderer
1995, Bruderer and Liechti 1998a).
For each night a tail wind component was calculated
as follows:
Tail wind component=[cos(b−v)] ·v
w
(3)
where b=basic direction, defined as mean track direc-
tion under low wind speed conditions (v
w
55ms
−1
)
where birds are expected to fully compensate for wind
drift (Bruderer et al. 1989), v=wind direction and
v
w
=wind speed (m s
−1
). At Falsterbo the basic direc-
tion of nocturnal migration in autumn 1998 was b=
225°. Negative values of tail wind correspond to head
wind.
Under the assumption that mean track direction only
varies according to wind, we can apply the ‘‘universal
measure of drift’’ (Alerstam 1976a) to describe the
amount of wind drift. The magnitude of drift was
estimated for each night as the slope coefficient b of the
relationship between mean track direction and the cor-
responding angle between mean track and mean head-
ing directions. A coefficient of b=0 means complete
compensation and b=1 full drift. An intermediate
value of b indicates partial drift compensation. In order
to correlate circular values in a linear function, we
restricted the analysed mean track directions to 180°
around the basic direction (135°–315°).
To compensate for the altitude-dependent detection
probability of birds aloft, mean track direction and
mean vector length were calculated according to the
increasing field of view with distance. In order to avoid
pseudoreplication in the analyses of track direction and
scatter in the course of the night (Fig. 8), we calculated
mean vectors based on mean values for each night.
Within single nights only time intervals (tenths of night)
with at least five birds were included, since means based
on less than five observations were considered as not
representative. Circular statistics (Rayleigh test, 95%
confidence interval, Watson-Williams test and a non-
parametric test for dispersion) were applied according
to Batschelet (1981). Chi-square and Kendall-ttests,
correlation analyses and analyses of multiple regression
(stepwise forward) were done with STATISTICA 5.0
(StatSoft, Inc. 1995) software package. For the multiple
regression analysis the requirement of normally dis-
tributed residuals was achieved by log-transforming the
dependent variable (mean MTR) according to Stahel
(1995). For the analysis of weather factors, originally
all weather variables recorded at Falsterbo, the calcu-
lated trends, and the tailwind component were included
as independent variables.
Results
Temporal patterns
Observations were carried out during 51 entire nights,
while in 17 nights observations were interrupted due to
occasional precipitation. Fifteen nights had to be ex-
cluded because of rain. In total, 17411 birds were
registered in 632.3 h of observation. The mean migra-
tion traffic rate per night fluctuated between 6 and 6618
242 JOURNAL OF AVIAN BIOLOGY 32:3 (2001)

Fig. 4. Seasonal pattern of mean MTR (birds km
−1
h
−1
) per
night. Nights without observations due to precipitation are
marked with grey bars.
Fig. 5. Nightly patterns of migratory intensities. The duration
of the night is limited by civil twilight at dusk and at dawn
(lines). The dots mark the median (50% of the nocturnal
migration has passed) and the T-bars show the time interval
between the 25% and 75% percentiles. The time scale corre-
sponds to local summertime (UTC+2 h).
birds km
−1
h
−1
(Fig. 4), with an average of 1319 birds
km
−1
h
−1
and a median of 566 birds km
−1
h
−1
for
those nights with observations. Assuming that missing
nights had a MTR of 0, the mean migration traffic rate
for the period between 7 August and 28 October was
1081 birds km
−1
h
−1
and the median 327 birds km
−1
h
−1
. Nightly mean MTRs were moderate in the first
half of the season and peaked in the second half of
September. In October the MTR declined considerably.
Wind speed, tailwind component, air pressure and
change in air pressure accounted for 66% of the vari-
ance in MTR (Table 1). Nightly mean MTR was high
in conditions of low wind speed, large tailwind compo-
nent, high and rising air pressure. Temperature and
temperature change, humidity, dew point temperature,
distance to the cloud base, and degree of cloud cover
had no significant influence on the numbers of birds
observed.
The median of the nightly passage (MTR) was
reached on average after 40% (SD915%) of the night’s
duration (Fig. 5). On average this corresponds to4h53
min (SD91 h 47 min) after the end of civil twilight.
The median was not correlated with the end of civil
twilight at dusk, neither for all nights (N=66, r=0.14,
p=0.28), nor for nights with intense migration only
(N=16, r=0.39, p =0.14).
The duration of the main passage (50% around the
median) in August and September decreased signifi-
cantly with increasing migratory intensity (N=45, r=
−0.50, pB0.001). But this correlation is not significant
when data from October are included (N=68, r=
−0.16, p=0.21). In October variability increased due
to low migratory intensities, partly caused by inclement
weather. The overall nocturnal pattern showed a pro-
nounced peak of migratory intensity in the first half of
the night and a prolonged decline towards the morning
(Fig. 6).
The comparison of three time intervals of the night
showed that during peak migratory intensity (tenths
2–5 of the night) birds flew higher than in the first
tenth (x
2
=315.6, df=6, pB0.001) and also higher
than in the remaining part of the night (tenths 6–10;
x
2
=328.3, df=6, pB0.001; tests based on the seven
height classes; Fig. 7).
Fig. 6. Nocturnal pattern of mean MTR (birds km
−1
h
−1
)
calculated from the data of the whole season. The error bars
indicate the interquartile range.
Table 1. Result of a multiple regression analysis (forward and
backward stepwise, type III) relating mean MTR per night
(log-transformed) with weather variables (R
2
=0.657,
F
(4,63)
=12.309, p0.001, S.E. of estimate=0.386). Mean
MTRs are shown in Fig. 4.
B (S.E. of B) t (53) p-level
Intercept 26.98 (6.09) 4.43 B0.001
−6.67−0.11 (0.02)Wind speed B0.001
0.06 (0.01) 6.11 B0.001Tailwind
component
Air pressure −0.02 (0.01) −3.87 B0.001
Trend of air B0.0010.03 (0.01) 4.78
pressure
JOURNAL OF AVIAN BIOLOGY 32:3 (2001) 243

Fig. 7. Distribution of flight altitudes in three time intervals.
The migratory intensity per height class is given as a propor-
tion of the mean MTR per time interval. A includes the first
tenth of the night; B tenths 2–5 of the night; C tenths 6–10 of
the night.
Fig. 8. Mean track direction during the course of the night.
The black symbols show the mean track directions calculated
based on every single track weighted by its contribution to the
MTR. The error bars indicate the 95% confidence interval
(Batschelet 1981). The open circles show the track direction
calculated based on nightly means. During the migratory peak
(tenth 2–5 of the night) tracks were significantly more concen-
trated and southerly oriented than in the rest of the night.
Flight directions and wind influence
The mean track direction of all birds including all
nights was 219°(N=17411, r=0.62, p B0.001; 99°,
95% confidence interval). The higher the flight altitude
was, the more southerly and more concentrated were
the track directions (N=7, Z=3.15, p =0.002;
Kendall; for both relationships) (Table 2).
In the first tenth of the night, tracks were more
westerly oriented than during the second tenth period
(Fig. 8; N=89, F=4.20, p B0.05; Watson-Williams
test). In the second half of the night (6–10 tenth of the
night) track directions gradually shifted from south-
westerly directions towards more westerly directions
(N=385, F=22.21, p B0.01; Watson-Williams test).
Birds recorded late at night (tenth 6–10 of the night)
were more scattered, mainly because of an increasing
tendency to fly towards W and NW during this period
(N=385, U=15944, p B0.001; nonparametric test for
dispersion).
We related the mean track direction to the mean
angle between track and heading direction for each
night in order to analyse the birds’response to wind
drift (Alerstam 1976a). In nights with high and low
migratory intensity, mean track directions were posi-
tively correlated with the angle between track and
heading at altitudes below and above 1000 m ASL
(linear regression for high intensity: B1000 m: N=17,
r=0.70, p=0.002; N=16, r =0.88, p B0.001, and for
low intensity: B1000 m: N =21, r =0.69, p B0.001;
\1000 m: N=18, r=0.79, p B0.001) (Fig. 9). In
nights with intermediate migratory activity the correla-
tion was positive in both flight altitude categories, but
not significant. Only for one of the significant correla-
tions, i.e. for intense migration at high altitude, the
slope of the regression line was significantly different
from 1, indicating partial compensation (see Fig. 9).
For the analysis of mean track directions in relation
to the coastline, only low flying birds (B600 m ASL),
i.e. height classes 7 and 6 were included, since the
influence of topographical structures have been shown
to decrease with altitude (Bruderer et al. 1989). Under
conditions of low wind speed (v
w
55ms
−1
) 51% of
the migratory movements were recorded in height
classes 6–10. The mean track directions of this selection
differed between winds from the left (v=44°to 226°;
a=231°,r=0.49, N =1536, 95% CI 912°) and from
Table 2. Mean track directions and MTR with respect to flight altitude as given for each size class of birds. Flight altitude was
estimated according to silhouette size (see Study site and methods). All directional distributions were significantly different from
random (pB0.001, Rayleigh test).
Size class 7 465 3 12
Estimated mean flight altitude (km) 2.11.91.51.10.70.50.3
Mean track direction (°) 228 227 226 219 209 201 199
0.92Mean vector length 0.45 0.65 0.66 0.75 0.80 0.87
136 108 52532 218MTR (birds km
−1
h
−1
) 190 167
3216 3241 17232517Number of tracks 1722 2098 2894
244 JOURNAL OF AVIAN BIOLOGY 32:3 (2001)

the right (v=226°to 44°;a=220°,r=0.57, N=826,
95% CI910°), though the 95% confidence intervals
overlap. The orientation of the coastline 100 m west of
the observation site (21°/201°) was not within the calcu-
lated 95% confidence intervals of the mean track direc-
tion, neither with winds from the left nor from the
right.
Discussion
Temporal pattern of migration
The migratory intensities recorded at Falsterbo re-
vealed large fluctuations between nights. These fluctua-
tions are most probably due to strong weather changes
that were especially pronounced in 1998 (L. Karlsson,
pers. comm.), but could also reflect the variation in
number of birds ready to migrate at different times of
the season (Richardson 1978). The mean seasonal MTR
of 1319 km
−1
h
−1
is the first estimate of nocturnal
migratory intensity in Scandinavia. Assigning zero val-
ues to times when no observations were possible due to
rain, we would come up with a mean MTR of roughly
1000 birds km
−1
h
−1
for the total observation period.
Comparisons with other sites will be necessary to verify
whether nocturnal migration is concentrated at Fal-
sterbo or not. The nightly mean MTRs show the depar-
ture of long-distance migrants in August, a peak of
migration in the second half of September, when long-
and short-distance migrants pass Falsterbo, and declin-
ing numbers of short-distance migrants in October, as
known from long-term ringing programs (Karlsson
1993). The number of nocturnal migrants observed
aloft correlated with the number of grounded birds
captured the following morning (Zehnder and Karlsson
2001). The seasonal pattern of nocturnal migration is
similar to the one found by Baumgartner (1997) in
southern Germany (see also Bruderer and Liechti
1998b). Baumgartner (1997) compared long-term trap-
ping numbers of birds at a ringing site in the Alps and
radar data of one season, emphasising that it is impor-
tant to account for seasonal effects when analysing
weather influences. We did not include seasonal effects
in the multiple regression analysis of weather factors on
migratory intensity. However, almost two thirds of the
variance in migratory intensity could be explained by
variations in wind (tailwind component and wind veloc-
ity) and air pressure (see above). Weak tail winds are
well known to be associated with heavy migratory
movements (for review see Richardson 1978). By select-
ing nights with favourable wind conditions for migra-
tion a bird may considerably increase its flight distance
covered on a given amount of fuel (Alerstam 1979,
Alerstam and Lindstro¨m 1990, Liechti and Bruderer
1998, A
,
kesson and Hedenstro¨m 2000). Air pressure is
often reported to correlate with migratory intensity,
particularly in North America (see Richardson 1978,
1990 for reviews). However, Richardson (1978) empha-
sises that air pressure is correlated with many other
weather factors. Falling and low air pressures generally
are associated with deteriorating weather not suitable
for migration and frequently involves rain. Rain to a
large extent prevents birds from initiating migration
(Parslow 1969). Indeed, we observed strong influences
of wind and increasing or high air pressure on the
intensity of nocturnal migration at Falsterbo in line
with the findings by Baumgartner (1997). High air
pressures may be taken as an indicator of a generally
good synoptic weather situation, while low pressure
may be taken as a substitute for the negative influence
of precipitation.
We observed that most of the nights with high migra-
tory intensity were followed by a night with relatively
low migratory intensity. This pattern indicates that
many nocturnal migrants that were prepared for migra-
Fig. 9. Nightly mean track
direction in relation to track
direction minus heading direction
of birds flying above and below
1000 m ASL, respectively. Nights
with intense migration are marked
with a dot, intermediate intensity
with an open circle and low
intensity with a cross. The slope of
the regression line is a measure of
the magnitude of drift or
compensation for drift (b=1
means full drift and b=0 complete
compensation) (Alerstam 1976a)
(Intense migration: B1000 m ASL,
b=0.9190.16 (95% confidence
interval), \1000 m ASL,
b=0.7190.07, solid lines; low
intensity migration: B1000 m
ASL, b=0.9090.15, \1000 m
ASL, b=0.9290.12, broken lines)
(see text for more statistics).
JOURNAL OF AVIAN BIOLOGY 32:3 (2001) 245

tion departed during the limited time periods with
favourable weather conditions (see above). Presumably
it took some time before newly arrived migrants from
further north were prepared to re-initiate migration. As
shown by several studies, the vast majority of migrants
depart in the first two hours after sunset (reviews in
Moore 1987, Kerlinger and Moore 1989, A
,
kesson et al.
1996a, 2001, Bruderer and Liechti 1999). At Falsterbo
the timing of the peak migratory passage was generally
later at night and varied between days. Hence, birds
most likely reached Falsterbo after flights of different
lengths (i.e. arriving from source areas at different
distances from Falsterbo). Since birds maintained their
headings irrespective of wind conditions we assume
that, depending on wind direction and wind speed,
birds from different stopover areas were observed at
Falsterbo. Winds additionally influence flight speeds,
and adverse weather conditions at the beginning of the
night may delay departure.
The majority of the birds apparently arrived at Fals-
terbo after several hours of flight. Estimating a bird’s
flight speed at 12 m s
−1
and by extrapolating the mean
track direction we can estimate that the majority of the
recorded migrants took off from the forested areas
more than 100 km to the NE of Falsterbo. The major-
ity of the migrants were seen during the first half of the
night, with relatively low migratory intensities in the
second half of the night. This indicates that birds
presumably begin to land already after approximately 3
to6hofflight. Such early landings of nocturnal
migrants have earlier been reported in radar studies
(e.g. Bruderer and Liechti 1998b). In correspondence
with this landing tendency, flight altitudes declined
gradually during the second half of the night. A similar
pattern has been observed at an inland site in southern
Germany (Bruderer and Liechti 1998b), and thus need
not indicate a landfall caused by the birds being con-
fronted with the sea.
Flight directions and wind drift
Nights with intense migratory activity showed mean
directions towards 225°at Falsterbo, which is similar to
the main migratory directions reported for other sites in
western Europe (e.g. 230°for southern Germany and
Switzerland, Bruderer et al. 1989; 220°at a coastal site
in southern Spain, Bruderer and Liechti 1998a). We
also observed a more southerly mean track direction at
higher altitudes which is in agreement with former
observations, and presumably could be explained by
the increasing wind speed with altitude. Since winds
from the west predominated throughout the season,
birds higher up were drifted more southwards than
birds at lower altitudes. Pseudodrift might account at
least for part of the change with altitude if birds with
more southerly preferred migratory directions actively
choose to fly at higher altitudes (cf. Alerstam 1976b).
The shift of the mean track direction during the night
from SSW to SW resulted from a higher proportion of
birds flying in westerly and north-westerly directions.
These directions (W and NW) allow a shorter sea
crossing from Falsterbo to Denmark (24 km between
the closest points). In most nights the coastlines of
Denmark towards W–SW are within sight of a bird
flying above Falsterbo. After some hours of flight birds
might be less prone to face a large sea crossing south-
wards due to the endogenous activity scheme or due to
depleted fat stores. They may then choose, instead, to
shorten the over-water flight by heading towards the
Danish islands. A similar shift of the mean track direc-
tion was observed at various sites along the Mediter-
ranean coast, where with the progress of the night birds
increasingly shifted from southerly off-shore directions
towards flying along the coastline (Fortin et al. 1999).
Reverse movements recorded by the infrared device
were rare (S. Zehnder, S. A
,
kesson, F. Liechti and B.
Bruderer unpubl.) and much fewer than expected from
an analysis of recoveries of birds ringed at this site
(A
,
kesson et al. 1996b) and recorded by ceilometer
observations (A
,
kesson 1995). Our observations indicate
that once the birds had left the coast they usually did
not return. This differs from relatively large sea-cross-
ings at the south coast of Spain and on the island of
Mallorca (Bruderer and Liechti 1998a) but conforms
with various North American studies (review in
Williams and Williams 1990), including very long over-
water flights such as from the southern tip of Florida
(Williams et al. 1977). Birds migrating at high altitudes
seem to engage in sea-crossings towards the closest land
on the opposite site of the sea (directed towards SW;
see Fig. 1), rather than flying in reverse directions to
stopover sites further inland in southern Sweden, as
was reported for migrants grounded and captured at
Falsterbo bird observatory (A
,
kesson et al. 1996b).
Migratory activity was significantly higher at low
wind speeds. Surprisingly, the birds’headings rather
than tracks showed a constant mean direction, indicat-
ing wind drift at low speeds, where full wind compensa-
tion could be expected (Alerstam 1976a, Liechti 1993).
Birds at low wind speeds apparently maintained their
preferred heading independently of wind direction and
at least partially tolerated wind drift. Constant heading
and lack of compensation for wind drift has been
reported for different places in North America (e.g.
Williams et al. 1977; review in Williams and Williams
1990). These observations might be explained by the
predictions of optimal wind compensation relative to
the distance to the migratory goal, since birds should
allow themselves to be drifted at the beginning of the
journey and over-compensate when they are close to
the goal (Alerstam 1979, Liechti 1995). For the major-
ity of the bird species recorded in this study the migra-
tory goal is relatively far away. Since bird species or
246 JOURNAL OF AVIAN BIOLOGY 32:3 (2001)

populations with different migratory goals might select
to migrate at nights with a particular wind condition
(Alerstam 1972, 1975, see also A
,
kesson and Heden-
stro¨m 2000) we cannot exclude that this pattern of wind
drift, at least partly, is caused by pseudodrift (Alerstam
1976a).
Nocturnal passerine migrants at low altitudes (10–
100 m ASL) have been shown to compensate for wind
drift at this particular site, by flying along the coastlines
of the Falsterbo peninsula (A
,
kesson 1993a). At the
particular coastline where our observations took place
(site 1 in A
,
kesson 1993a), the wind compensation was
less marked. In the present study, we failed to demon-
strate any leading-line effect. Obvious leading-line ef-
fects have been found along rivers (Bingman et al.
1982) and mountain ranges running close to the pre-
ferred flight direction (Bruderer and Liechti 1990 and
references therein). Since the birds allow themselves to
be drifted, we assume that for the majority of nocturnal
migrants flying above 100 m AGL the local coastline
does not serve as a leading line. This is in line with
studies at continental coastlines and oceanic islands,
where birds have been observed moving on a broad-
front without deviations (Williams and Williams 1990).
Acknowledgements –We are very grateful to the staff of
Falsterbo Bird Observatory who kindly let us use their facili-
ties and gave practical support during field work. Many
thanks to Lennart Karlsson and Go¨ran Walinder for provid-
ing and introducing us to the weather data. We thank Katia
Hueso Kortekaas for analysing part of the video tapes and
Lorenz Gygax for statistical advice. Dieter Peter deserves
credit for developing the passive infrared method. He was
constantly present for further advice. Hardy Brun designed a
computer program to calculate cumulative frequencies. Many
thanks to Thomas Alerstam for fruitful discussions, particu-
larly about wind drift. Otto Holzgang, Matthias Kestenholz
and Dieter Peter read an earlier draft of the manuscript. H.-U.
Reyer made many valuable suggestions. We acknowledge ad-
ditional suggestions made by the two reviewers, S. A. Gau-
threaux and T. C. Williams. This study was financially
supported by the Swiss Ornithological Institute and a grant
from the Swedish Natural Science Research Council to S.
A
,
kesson. This is report No. 202 from Falsterbo Bird
Observatory.
References
Ahle´n, I. 1997. Migratory behaviour of bats at south Swedish
coasts. –Z. Sa¨ugetierk. 62: 375–380.
A
,
kesson, S. 1993a. Coastal migration and wind drift compen-
sation in nocturnal passerine migrants. –Ornis Scand. 24:
87–94.
A
,
kesson, S. 1993b. Effect of geomagnetic field on orientation
of the marsh warbler Acrocephalus palustris, in Sweden and
Kenya. –Anim. Behav. 46: 1157–1167.
A
,
kesson, S. 1994. Comparative orientation experiments with
different species of long-distance migrants: effect of mag-
netic field manipulation. –Anim. Behav. 48: 1379–1393.
A
,
kesson, S. 1995. Avian migratory orientation: geographic
and geomagnetic effects. –PhD thesis, University of Lund,
Sweden.
A
,
kesson, S. and Ba¨ckman, J. 1999. Orientation in pied
flycatchers: the relative importance of magnetic and visual
information at dusk. –Anim. Behav. 57: 819–828.
A
,
kesson, S. and Hedenstro¨m, A. 2000. Wind selectivity of
migratory departures in birds. –Behav. Ecol. Sociobiol.
47: 140–144.
A
,
kesson, S., Alerstam, T. and Hedenstro¨m, A. 1996a. Flight
initiation of nocturnal passerine migrants in relation to
celestial conditions at twilight. –J. Avian Biol. 27: 95–102.
A
,
kesson, S., Karlsson, L., Walinder, G. and Alerstam, T.
1996b. Bimodal orientation and the occurrence of tempo-
rary reverse bird migration during autumn in south Scandi-
navia. –Behav. Ecol. Sociobiol. 38: 293–302.
A
,
kesson, S., Walinder, G., Karlsson, L. and Ehnbom, S. 2001.
Reed warbler orientation: initiation of nocturnal migratory
flights in relation to visibility of celestial cues at dusk. –
Anim. Behav. 61: 181–189.
Alerstam, T. 1972. Nocturnal bird migration in Ska˚ne, Swe-
den, as recorded by radar in autumn 1971. –Ornis Scand.
3: 141–151.
Alerstam, T. 1975. Redwing (Turdus iliacus) migration to-
wards southeast over southern Sweden. –Vogelwarte 28:
2–17.
Alerstam, T. 1976a. Bird migration in relation to wind and
topography. –PhD thesis, Univ. of Lund, Sweden.
Alerstam, T. 1976b. Nocturnal migration of thrushes (Turdus
spp.) in southern Sweden. –Oikos 27: 457–475.
Alerstam, T. 1978. Analysis and a theory of visible bird
migration. –Oikos 30: 273–349.
Alerstam, T. 1979. Optimal use of wind by migrating birds:
combined drift and overcompensation. –J. theor. Biol. 79:
341–353.
Alerstam, T. 1990. Bird Migration. –Cambridge University
Press, Cambridge.
Alerstam, T. and Lindstro¨m, A
,
. 1990. Optimal bird migration:
the relative importance of time, energy, and safety. –In:
Gwinner, E. (ed.). Bird Migration: Physiology and Eco-
physiology. –Springer-Verlag, Berlin, pp. 331–351.
Alerstam, T. and Ulfstrand, S. 1972. Radar and field observa-
tions of diurnal bird migration in south Sweden, autumn
1971. –Ornis Scand. 3: 99–139.
Alerstam, T. and Ulfstrand, S. 1975. Diurnal migration of
passerine birds over south Sweden in relation to wind
direction and topography. –Ornis Scand. 6: 135–149.
Batschelet, E. 1981. Circular Statistics in Biology. –Academic
Press, London.
Baumgartner, M. 1997. Wetterabha¨ngigkeit des na¨chtlichen
Vogelzuges im Herbst u¨ber Su¨ddeutschland. –PhD thesis,
Schweizerische Vogelwarte Sempach, Switzerland.
Bingman, V. P., Able, K. P. and Kerlinger, P. 1982. Wind
drift, compensation, and the use of landmarks by noctur-
nal bird migration. –Anim. Behav. 30: 49–53.
Bloch, R., Bruderer, B. and Steiner, P. 1981. Flugverhalten
na¨chtlich ziehender Vo¨gel –Radardaten u¨ber den Zug
verschiedener Vogeltypen auf einem Alpenpass. –Vogel-
warte 31: 119–149.
Bruderer, B. 1997. Migratory directions of birds under the
influence of wind and topography. RIN 1997: Orientation
and Navigation –Birds, Humans and Other Animals. –
The Royal Institute of Navigation, Oxford 27: 1–10.
Bruderer, B. and Liechti, F. 1990. Richtungsverhalten
nachtziehender Vo¨gel in Su¨ddeutschland und der Schweiz
unter besonderer Beru¨cksichtigung des Windeinflusses. –
Ornithol. Beob. 87: 271–293.
Bruderer, B. and Liechti, F. 1994. Quantification of bird
migration –different means compared. –Proc. Bird Strike
Committee Europe 22: 243–254.
Bruderer, B. and Liechti, F. 1998a. Flight behaviour of noc-
turnally migrating birds in coastal areas –crossing or
coasting. –J. Avian Biol. 29: 499–507.
Bruderer, B. and Liechti, F. 1998b. Intensita¨t, Ho¨he und
Richtung von Tag- und Nachtzug im Herbst u¨ber Su¨dwest-
deutschland. –Ornithol. Beob. 95: 113–128.
Bruderer, B. and Liechti, F. 1999. Bird migration across the
Mediterranean. –In: Adams, N. and Slotow, R. (eds).
Proc. 22 Int. Ornithol. Congr, Durban, BirdLife South
Africa, Johannesburg, pp. 1983–1999. (CD-ROM.)
JOURNAL OF AVIAN BIOLOGY 32:3 (2001) 247

Bruderer, B., Liechti, F. and Erich, D. 1989. Radarbeobachtun-
gen u¨ber den herbstlichen Vogelzug in Su¨ddeutschland. –
Vogel und Luftverkehr 9: 174–194.
Fortin, D., Liechti, F. and Bruderer, B. 1999. Variation in the
nocturnal flight behaviour of migratory birds along the
northwest coast of the Mediterranean Sea. –Ibis 141:
480–488.
Karlsson, L. (ed.). 1993. Birds at Falsterbo. –Anser, suppl. 33.
Kerlinger, P. and Moore, F. R. 1989. Atmospheric structure
and avian migration. –In: Power, D. (ed.). Current Or-
nithology. Vol. 6. Plenum, New York, pp. 109–142.
Koch, J. C. 1934. Vogelzug unter Einfluss von Leitlinie und
Windrichtung. –Vogelzug 5: 45–52.
Liechti, F. 1993. Na¨chtlicher Vogelzug im Herbst u¨ber Su¨d-
deutschland: Winddrift und Kompensation. –J. Ornithol.
134: 373–404.
Liechti, F. 1995. Modelling optimal heading and airspeed of
migrating birds in relation to energy expenditure and wind
influence. –J. Avian Biol. 26: 330–336.
Liechti, F. and Bruderer, B. 1995. Direction, speed and compo-
sition of nocturnal bird migration in the south of Israel. –
Israel J. Zool. 43: 501–515.
Liechti, F. and Bruderer, B. 1998. The relevance of wind for
optimal migration theory. –J. Avian Biol. 29: 561–568.
Liechti, F., Bruderer, B. and Paproth, H. 1995. Quantification
of nocturnal bird migration by moonwatching: comparison
with radar and infrared observations. –J. Field Ornithol.
66: 457–468.
Liechti, F., Peter, D., Lardelli, R. and Bruderer, B. 1996. Die
Alpen, ein Hindernis im na¨chtlichen Breitfrontenzug –eine
grossra¨umige U
8
bersicht nach Mondbeobachtungen. –J.
Ornithol. 137: 337–356.
Lowery, G. H. 1951. A quantitative study of the nocturnal
migration of birds. –Univ. Kansas Publ., Mus. of Nat.
Hist. 3: 361–472.
Lowery, G. H. and Newman, R. J. 1955. Direct studies of
nocturnal bird migration. –In: Wolfson, A. (ed.). Recent
Studies in Avian Biology. Univ. of Illinois Press, Urbana,
pp. 238–263.
Moore, F. R. 1987. Sunset and the orientation of behaviour of
migrating birds. –Biol. Rev. 62: 65–86.
Parslow, J. L. F. 1969. The migration of passerine night
migrants across the English Channel studied by radar. –Ibis
111: 48–79.
Richardson, W. J. 1978. Timing and amount of bird migration
in relation to weather: a review. –Oikos 30: 224–274.
Richardson, W. J. 1990. Timing of bird migration in relation
to weather: Updated review. –In: Gwinner E. (ed.). Bird
Migration: Physiology and Ecophysiology, Springer-Verlag,
Berlin, pp. 78–101.
Richardson, W. J. 1991. Wind and orientation of migrating
birds: A Review. –In: Berthold, P. (ed.). Orientation in
Birds, Birkha¨user Verlag, Basel, pp. 226–249.
Sandberg, R., Pettersson, J. and Alerstam, T. 1988. Why do
robins, Erithacus rubecula, captured at two nearby stop-over
sites orient differently? –Anim. Behav. 36: 877–887.
Stahel, W. A. 1995. Statistische Datenanalyse –eine Ein-
fu¨hrung fu¨r Naturwissenschaftler. –Verlag Vieweg,
Braunschweig.
StatSoft, Inc. 1995. Statistica. –Tulsa, Oklahoma.
Williams, T. C. and Williams, J. M. 1990. The orientation of
transoceanic migrations. –In: Gwinner E. (ed.). Bird
Migration: Physiology and Ecophysiology, Springer-Verlag,
Berlin, pp. 7–21.
Williams, T. C., Berkeley, P. and Harris, V. 1977. Autumnal
bird migration over Miami studied by radar: a possible test
of the wind drift hypothesis. –Bird Banding 48: 1–10.
Zehnder, S. and Karlsson, L. Do ringing numbers reflect true
migratory activity of nocturnal migrants? –J. Ornithol. In
press.
(Recei6ed
23
May
2000
,re6ised
6
December
2000
,accepted
22
December
2000
.)
248 JOURNAL OF AVIAN BIOLOGY 32:3 (2001)