Timing of breeding area arrival is crucial for
migratory birds to profit from the summer peak in
food availability, thereby optimizing their reproductive output (Both et al. 2006). As consistent patterns
of earlier arrival to their northern Europe breeding
areas have been found ( Thorup et al. 2007), a
number of questions arise: what factors drive these
changes and which cues do the migrants use to
determine optimal arrival? In fact, we know little
about which climatic variables affect migrants during
different seasons and phases of the migration cycle
(Stenseth & Mysterud 2005).
Timing of spring migration can be affected by
environmental conditions in two ways. The first
occurs prior to departure from wintering areas where
ecological conditions can affect individuals’ physiology, e.g. body condition (Ottosson et al. 2005). The
second occurs during the migration period, where
improved ecological conditions at stopover sites with
an abundance of food can reduce the time required to
replenish fuel stores (Ahola et al. 2004).
Recently, Jonzén et al. (2006) showed that longdistance migrants have advanced their passage through
Italy, making the journey earlier in the season, and
argued that this trend is the result of a climate-driven
evolutionary change. However, Both (2007) argued
that this trend might be caused by improved environmental conditions in the wintering areas, or en route,
while Gienapp et al. (2007) found no evidence to
support or disprove either an evolutionary or plastic
response. Our objective is to assess the average relationship between timing of migration at different latitudes
for several species of long-distance Palaearctic migrants
in the eastern and the central Mediterranean flyways,
and to analyse changes in the duration of migratory
periods relative to environmental conditions, both in
wintering areas and en route towards breeding areas.
Biol. Lett. (2008) 4, 685–688
doi:10.1098/rsbl.2008.0290
Published online 12 August 2008
Global change biology
Avian migrants adjust
migration in response to
environmental conditions
en route
Anders P. Tøttrup1,*, Kasper Thorup1,2,
Kalle Rainio3, Reuven Yosef 4, Esa Lehikoinen3
and Carsten Rahbek1
1
Center for Macroecology and Evolution, Department of Biology, and
Zoological Museum, University of Copenhagen, Universitetsparken 15,
2100 Copenhagen, Denmark
3
Section of Ecology, Department of Biology, University of Turku,
20014 Turku, Finland
4
International Birding and Research Centre, PO Box 774,
Eilat 88106, Israel
*Author for correspondence ([email protected]).
2
The onset of migration in birds is assumed to
be primarily under endogenous control in longdistance migrants. Recently, climate changes
appear to have been driving a rapid change in
breeding area arrival. However, little is known
about the climatic factors affecting migratory
birds during the migration cycle, or whether
recently reported phenological changes are
caused by plastic behavioural responses or
evolutionary change. Here, we investigate how
environmental conditions in the wintering areas
as well as en route towards breeding areas affect
timing of migration. Using data from 1984 to
2004 covering the entire migration period every
year from observatories located in the Middle
East and northern Europe, we show that passage
of the Sahara Desert is delayed and correlated
with improved conditions in the wintering areas.
By contrast, migrants travel more rapidly
through Europe, and adjust their breeding area
arrival time in response to improved environmental conditions en route. Previous studies
have reported opposing results from a different
migration route through the Mediterranean
region (Italy). We argue that the simplest explanation for different phenological patterns at
different latitudes and between migratory routes
appears to be phenotypic responses to spatial
variability in conditions en route.
2. MATERIAL AND METHODS
We used standardized ringing data on five long-distance migratory
species trapped en route from African wintering areas through the
Middle East (Israel) and just before arrival at their breeding areas
in northern Europe (Finland) during the period 1984–2004.
Timing of migration was analysed for three population measures
(first 5, 50 and 95% of the spring total), subsequently referred to as
‘migration phases’. Migration duration is defined as the difference
in passage dates between the Middle East and northern Europe for
each migration phase.
We used the monthly normalized difference vegetation index
(NDVI ) for the period 1984–2000 as a proxy measure for the actual
ecological conditions (e.g. food availability; Pettorelli et al. 2005) in
the species-specific wintering areas ( Wisz et al. 2007) prior to
departure (February/March) and the area in Europe traversed prior
to arrival at the breeding areas (April/May; figure 1). The African
spring NDVI and the European spring NDVI were then related to
our phenological results in the Middle East and northern Europe,
respectively (electronic supplementary material, methods).
We conducted two complementary analyses that have similar
findings, thus reported together in §3. First, we fitted models for
each species and migration phase, estimating changes in the timing
of migration, the effect of NDVI on the timing of migration as well
as the effect of NDVI on migration duration through Europe.
Second, we fitted linear mixed models including all independent
variables as fixed variables and their interactions as well as the three
phases of migration as a repeated factor. Our data include
observations of three phases of migration over 21 years from five
species at two locations, in total 630 observations (electronic
supplementary material, methods).
Finally, we relate our results and results published by Jonzén
et al. (2006) to NDVI in the eastern and western flyways between
Africa and Europe, to investigate whether migrants following
the two flyways experienced differently changing NDVI over time.
This included comparing changes in: (i) African winter NDVI
Keywords: birds; migration; phenology; climate
change; normalized difference vegetation index
1. INTRODUCTION
Migration phenology in long-distance avian migrants
is assumed to be controlled by endogenous rhythms
(Berthold 1996) with some degree of plasticity in the
migratory programme (Biebach et al. 1986). In the
Northern Hemisphere, spring temperatures have
increased during the past decades affecting, among
others, species’ physiology, breeding ranges and
phenology (Walther et al. 2002).
Electronic supplementary material is available at http://dx.doi.org/
10.1098/rsbl.2008.0290 or via http://journals.royalsociety.org.
Received 27 May 2008
Accepted 22 July 2008
685
This journal is q 2008 The Royal Society
686
A. P. Tøttrup et al.
Migration phenology in avian migrants
western
flyway
eastern
flyway
Europe
en route
Fi
sub-Saharan
Africa
wintering area
Is
northern
Africa
en route
It
phenology
later passage
vegetation
improved conditions
earlier passage
deteriorated conditions
Figure 1. Schematic of changes in the timing of migration for five long-distance migratory birds during the period
1984–2004 (red boxes) and changes in environmental conditions (NDVI ) for the period 1984–2000 (green boxes) in the
wintering areas and en route for the two major flyways (blue arrows) between Africa and Europe. Green areas were analysed
for environmental conditions (NDVI ) and the three ringing stations are indicated (Fi, Finland; It, Italy; Is, Israel).
Phenological data on the western flyway are from Jonzén et al. (2006).
(December–February) over time in the species-specific wintering
areas in western and eastern Africa (electronic supplementary
material, figure 1), (ii) NDVI along the central North African
and northeastern African flyways (March), and (iii) European
spring NDVI between western and eastern European flyways
(April; figure 1).
3. RESULTS
For the five long-distance migratory birds, we found
different trends in the timing of migration at different
latitudes (interaction term; latitude!year: p!0.0001,
FZ16.3, d.f.Z1). Figure 2a shows delayed passage
through the Middle East when compared with little
phenological change in northern Europe. The
difference was most clear for the portion of the
populations that arrive early, apparently decreasing
towards no overall difference between latitudes for the
portion of the populations that arrive late (latitude!
year!migration phase: pZ0.004, FZ5.52, d.f.Z2).
The species showed different trends in the timing of
migration (year!species: pZ0.004, FZ3.93, d.f.Z4),
most likely caused by different migration strategies.
Overall, NDVI affects the timing of migration
differently at different latitudes (latitude!NDVI:
pZ0.006, FZ7.76, d.f.Z1). Figure 2b shows that, in
all phases of migration, high African spring NDVI
delays passage, and for the first phases of migration,
the European spring NDVI correlated with earlier
passage in northern Europe. When included in the
same model, both year and NDVI correlated well
with the phenological patterns, suggesting a causal
Biol. Lett. (2008)
relationship with spring advancement. Furthermore,
advanced European spring NDVI seems to cause a
shorter migration duration in the first phases of
migration (5 and 50%; figure 2c) and the overall
analysis indicates more rapid migration in years
with high European spring NDVI (NDVI: pZ0.024,
FZ5.21, d.f.Z1; electronic supplementary material,
table 1). Excluding whitethroat from the analyses, the
species showing the most westerly ring recoveries
(electronic supplementary material, figure 2) did not
change the results.
The African winter NDVI increased significantly
in both western and eastern Africa (west: 0.013G
0.001 unit NDVI per year; east: 0.007G0.002 unit
NDVI per year). In central and northeastern North
Africa, NDVI showed no change over time (central:
pZ0.37, tZ0.93, R 2Z0.05; northeastern: pZ0.60,
tZK0.52, R 2Z0.02). Finally, the European spring
NDVI increased equally throughout the period in
western Europe (slope: 0.005 unit NDVI per year,
R 2Z0.28, pZ0.02) and eastern Europe (slope:
0.006 unit NDVI per year, R 2Z0.32, pZ0.01;
slopes not significantly different: pO0.1, tZ1.58,
d.f.Z34; figure 1).
4. DISCUSSION
During a period of improving conditions in the
wintering areas, five species of long-distance
migratory birds showed overall delayed spring passage
of the Sahara Desert. This indicates a delayed onset
Migration phenology in avian migrants A. P. Tøttrup et al.
(i)
(ii)
1.0
0.8
0.6
0.4
0.2
0
– 0.2
– 0.4
– 0.6
– 0.8
(b)
200
effect of NDVI
on passage
(days per unit NDVI)
change over time
(d yr –1)
(a)
160
120
80
40
0
– 40
– 80
effect of European spring
NDVI on migration duration
(days per unit NDVI)
(c)
140
100
60
20
–20
– 60
–100
–140
–180
–220
5% 50%
95%
5% 50% 95%
Figure 2. Species-specific changes in the timing of three
phases of spring migration (first 5, 50 and 95% of the
spring total) in five species of long-distance migratory birds
(lesser whitethroat (triangles), whitethroat (squares), blackcap (circles), chiffchaff (diamonds) and willow warbler
(plus symbols)) when passing the Middle East (filled
symbols) and northern Europe (empty symbols) analysed
for: (a) change in passage (d yrK1; 1984–2004); (b) relationship between passage in the Middle East and African spring
NDVI as well as passage in northern Europe and European
spring NDVI (days per unit NDVI; 1984–2000: positive
values indicate later passage); and (c) relationship between
change in migration duration and European spring NDVI
(days per unit NDVI; 1984–2000: negative values indicate
reduced migration duration). The five slope estimates for
each phase of migration are based on separate model fits
and are hence independent. (i) Individual estimates.
(ii) Mean values, where error bars are 95% confidence
intervals based onPthe summed sampling variances of each
species estimate, ð varðxi ÞÞ !ð1=5Þ2 .
of migration. Overall, trends in the timing of
migration in the Middle East were different from the
trends in northern Europe, commensurate with a
reduced migration period, where advantageous conditions reduce travel times during the second part of
their migration. These phenological patterns indicate
a high degree of phenotypic plasticity in the timing of
migration through Europe, with migrants adjusting
Biol. Lett. (2008)
687
their timing of migration to the actual conditions en
route and thereby optimizing their time of arrival at
the breeding areas.
In contrast to our results, Jonzén et al. (2006)
described an advanced passage of long-distance
migrants through Italy during approximately the same
time period covered by this study. These contradictory
results may have been caused by: (i) improved conditions along the western flyway making the birds
migrate faster when compared with the eastern flyway,
(ii) a difference in the latitude between the ringing
stations: the Italy data may reflect a response to
conditions in Europe, and (iii) monitoring of different
populations. At least during autumn migration, North
European migrants seem to circumvent the ecological
barriers of the Sahara Desert and the Mediterranean
Sea as opposed to crossing them (Fransson et al.
2005). Furthermore, Pilastro et al. (1998) observed
that some species of migratory songbirds trapped in
Italy had circumvented the Mediterranean Sea. This
implies that a large proportion of birds trapped in Italy
may belong to southern European populations. Hence,
prior to being trapped, the birds had traversed southwestern Europe where environmental conditions
improved in contrast to the central North Africa where
conditions were more constant (this study).
We found small advances in arrival to northeast
Europe, which supports the conclusions by Both et al.
(2004). Contrary to Gordo et al. (2005), who argued
that improved winter conditions cause earlier breeding area arrival, we found delayed initiation of
migration with improving winter area conditions.
Hence, migrants may benefit from improved conditions in the wintering areas by optimizing the
chance of safely passing the Sahara Desert. With
delayed onset of migration and advanced conditions
en route, the migratory birds have to speed up
migration through Europe to take advantage of a
climate-induced earlier spring. This supports the
study by Marra et al. (2005) showing advanced
migration with higher spring temperatures en route
through North America. As most birds included in
our data are second-year birds (electronic supplementary material, table 2) and hence on their first
northward migration, it seems unlikely that the
observed patterns result from birds reacting to previously experienced favourable conditions.
Data on passing migratory birds are most likely to
be from a mixture of populations of different origin,
and differences between locations may increase with
distance. Hence, different responses by different
populations may cause the patterns observed in this
and former studies. Also, local weather conditions
may cause between-species correlations in data on
passing migratory birds (Knape et al. in press),
potentially restricting conclusions drawn from phenological studies. Furthermore, different species and
even individuals may follow different decision rules
(Bowlin et al. 2005) and may not base their decisions
during flight and stopover on the progression of
vegetation, but rather follow simpler decision rules.
Also, using average values of NDVI from large areas
can be problematic as NDVI may vary considerably
and may even indicate different conditions, e.g.
688
A. P. Tøttrup et al.
Migration phenology in avian migrants
food availability in Africa and spring advancement
in Europe. However, adjustment of migration in
response to environmental conditions en route combined with the observed large variation in phenological patterns indicate that at least some degree of
phenotypic plasticity control the timing of breeding
area arrival. It seems unlikely that the relatively finescale adjustment along the migratory route could be
caused by microevolutionary changes. Yet, resolving
this question requires analyses of phenology at the
onset of spring migration and studies of evolutionary
changes within populations.
We thank the International Birding and Research Center,
Eilat and Jurmo bird observatory for collecting the data, as
well as the Finnish Museum of Natural History for ring
recovery data, and Niclas Jonzén, Andreas Lindén, Torbjørn
Ergon, Christiaan Both and two anonymous reviewers for
their helpful comments. Danish NSF ( J. no. 21-03-0221)
supported the research.
Ahola, M., Laaksonen, T., Sippola, K., Eeva, T., Rainio, K.
& Lehikoinen, E. 2004 Variation in climate warming
along the migration route uncouples arrival and breeding
dates. Glob. Change Biol. 10, 1610–1617. (doi:10.1111/
j.1365-2486.2004.00823.x)
Berthold, P. 1996 Control of bird migration. London, UK:
Chapman and Hall.
Biebach, H., Friedrich, W. & Heine, G. 1986 Interaction of
body mass, fat, foraging and stopover period in transSahara migrating passerine birds. Oecologia 69, 370–379.
(doi:10.1007/BF00377059)
Both, C. 2007 Comment on “Rapid advance of spring
arrival dates in long-distance migratory birds”. Science
315, 598. (doi:10.1126/science.1136148)
Both, C. et al. 2004 Large-scale geographical variation
confirms that climate change causes birds to lay earlier.
Proc. R. Soc. B 271, 1657–1662. (doi:10.1098/rspb.
2004.2770)
Both, C., Bouwhuis, S., Lessells, C. M. & Visser, M. E.
2006 Climate change and population declines in a longdistance migratory bird. Nature 441, 81–83. (doi:10.
1038/nature04539)
Bowlin, M. S., Cochran, W. W. & Wikelski, M. C. 2005
Biotelemetry of New World thrushes during migration:
physiology, energetics and orientation in the wild. Integr.
Comp. Biol. 45, 295–304. (doi:10.1093/icb/45.2.295)
Fransson, T., Jakobsson, S. & Kullberg, C. 2005 Nonrandom distribution of ring recoveries from transSaharan migrants indicates species-specific stopover
areas. J. Avian Biol. 36, 6–11. (doi:10.1111/j.0908-8857.
2005.03471.x)
Biol. Lett. (2008)
Gienapp, P., Leimu, R. & Merilä, J. 2007 Responses to
climate change in avian migration time—microevolution
versus phenotypic plasticity. Clim. Res. 35, 25–35.
(doi:10.3354/cr00712)
Gordo, O., Brotons, L., Ferrer, X. & Comas, P. 2005 Do
changes in climate patterns in wintering areas affect the
timing of the spring arrival of trans-Saharan migrant
birds? Glob. Change Biol. 11, 12–21. (doi:10.1111/
j.1365-2486.2004.00875.x)
Jonzén, N. et al. 2006 Rapid advance of spring arrival dates
in long-distance migratory birds. Science 312,
1959–1961. (doi:10.1126/science.1126119)
Knape, J., Jonzén, N., Sköld, M. & Sokolov, L. In press.
Multivariate state space modelling of bird migration
count data. In Modeling demographic processes in marked
populations. Environmental and ecological statistics, vol. 3
(eds D. L. Thomson, E. G. Cooch & M. J. Conroy).
Berlin, Germany: Springer.
Marra, P. P., Francis, C. M., Mulvihill, R. S. & Moore,
F. R. 2005 The influence of climate on the timing and
rate of spring bird migration. Oecologia 142, 307–315.
(doi:10.1007/s00442-004-1725-x)
Ottosson, U., Waldenström, J., Hjort, C. & McGregor, R.
2005 Garden warbler Sylvia borin migration in sub-Saharan
West Africa: phenology and body mass changes. Ibis 147,
750–757. (doi:10.1111/j.1474-919X.2005.00460.x)
Pettorelli, N., Vik, J. O., Mysterud, A., Gaillard, J.-M.,
Tucker, C. J. & Stenseth, N. Chr. 2005 Using the
satellite-derived NDVI to assess ecological responses to
environmental change. Trends Ecol. Evol. 20, 503–510.
(doi:10.1016/j.tree.2005.05.011)
Pilastro, A., Macchio, S., Massi, A., Montemaggiori, A. &
Spina, F. 1998 Spring migratory routes of eight transSaharan passerines through the central and western
Mediterranean; results from a network of insular and
coastal ringing sites. Ibis 140, 591–598. (doi:10.1111/
j.1474-919X.1998.tb04704.x)
Stenseth, N. C. & Mysterud, A. 2005 Weather packages:
finding the right scale and composition of climate in
ecology. J. Anim. Ecol. 74, 1195–1198. (doi:10.1111/
j.1365-2656.2005.01005.x)
Thorup, K., Tøttrup, A. P. & Rahbek, C. 2007 Patterns of
phenological changes in migratory birds. Oecologia 151,
697–703. (doi:10.1007/s00442-006-0608-8)
Walther, G. R., Post, E., Convey, P., Menzel, A., Parmesan,
C., Beebee, T. J., Fromentin, J. M., Hoegh-Guldberg, O. &
Bairlein, F. 2002 Ecological responses to recent climate
change. Nature 416, 389–395. (doi:10.1038/416389a)
Wisz, M. S., Walther, B. A. & Rahbek, C. 2007 Using
potential distributions to explore determinants of
western Palaearctic migratory songbird species richness
in sub-Saharan Africa. J. Biogeogr. 34, 828–841. (doi:10.
1111/j.1365-2699.2006.01661.x)
Supplementary figure 1 (A-E). Species specific wintering areas according to Wisz et al. (2007) for
western Africa (red) and eastern Africa (green). For each of these areas, we computed monthly NDVI
maximum value composite for the period 1984 to 2000 as a proxy measure for the actual environmental
conditions during the winter season (mean NDVI values for the period December to February).
(A) Lesser Whitethroat Sylvia curruca
D) Chiffchaff Phylloscopus collybita
(B) Whitethroat Sylvia communis
(F) Willow Warbler Phylloscopus trochilus
(C) Blackcap Sylvia atricapilla
Supplementary figure 2. Ring-recoveries of birds ringed at Eilat (29ºN 34ºE; Israel) and in Finland
during the years 1984 to 2004 and 1971 to 2001, respectively. Together with recovery maps in Zink
(1973) this map confirms that the populations breeding in eastern Fennoscandia migrate via the
Middle East to and from their wintering areas in tropical Africa. Circles indicate recoveries of all
birds ringed in Eilat and diamonds are all birds ringed in Finland and recovered abroad during
spring migration through Europe. Blue symbols represent Lesser Whitethroats Sylvia curruca, red
are Whitethroats Sylvia communis, black are Blackcaps Sylvia atricapilla, green are Chiffchaffs
Phylloscopus collybita and yellow symbols are Willow Warblers Phylloscopus trochilus.
Supplementary table 1. (A) The effect on timing of migration when including ‘Latitude’ (Eilat,
Middle East and Jurmo, northern Europe), ‘Year’ (1984 – 2004) and ’Species’ (see the text) and all
three ’Migration phases’ (first 5%, 50% and 95% of the total number of trapped or observed
individuals) in a statistical model. We show opposing trends at different latitudes (significant
latitude * year interaction) with delayed passage through the Middle East, and advanced arrival in
northern Europe (see figure 2A). The results for the two interaction terms: “Latitude*Year” and
“Year*Species” reported in the text are slightly different from the results reported below as the
higher order significant interaction term “Latitude*Year*Migration phase” has been removed from
the models reported in the text. (B) The effect on timing of migration in the Middle East and
northern Europe including latitude, year, species, and all three migration phases, as well as African
spring NDVI (in relation to Middle East passage) and European spring NDVI (in relation to
northern Europe passage) in a statistical model. We show opposing effects of NDVI at different
latitudes (significant latitude * NDVI interaction) (see figure 2B). (C) Effect on migration duration
through Europe (number of days between the Middle East and northern Europe for the 50%
migration phase) when including species, European spring NDVI and all three migration phases in a
statistical model. We show an overall significant effect of NDVI on migration duration, with better
conditions decreasing migration duration. All models were fitted as statistical models including all
independent variables and their interactions, using linear mixed models (proc MIXED; SAS 2000)
and with migration phase as the repeated factor. We performed backward elimination procedures
removing all insignificant variables except when higher-level interaction was significant.
A
B
Parameter
Timing of migration
Latitude
Year
Species
Migration phase
Latitude * Year
Latitude * Species
Latitude * Migration phase
Year * Species
Year * Migration phase
Species * Migration phase
Latitude * Year * Migration phase
Error
Timing of migration
Latitude
Year
Migration phase
NDVI
Species
Latitude * Migration phase
Latitude * Year
Latitude * NDVI
Latitude * Species
Year * Migration phase
Migration phase * NDVI
Migration phase * Species
Year * NDVI
df
F
P
1
1
4
2
1
4
2
4
2
8
2
550
16.13
4.58
3.89
0.51
16.12
0.01
5.52
3.89
0.51
0.00
5.52
< 0.0001
0.0329
0.0040
0.6000
< 0.0001
0.9999
0.0042
0.0040
0.5999
1.0000
0.0042
1
1
2
1
4
2
1
1
4
2
2
8
1
0.21
3.68
0.72
0.98
1.96
3.77
0.21
7.76
0.00
0.72
0.89
0.00
0.99
0.6443
0.0559
0.4874
0.3217
0.1000
0.0238
0.6439
0.0056
1.0000
0.4874
0.4107
1.0000
0.3207
C
Year * Species
Latitude * Year * Migration phase
Error
2
2
441
3.77
3.77
0.0999
0.0237
Migration duration
Species
Migration phase
Species * Migration phase
NDVI
Error
4
2
8
1
188
87.08
27.56
6.64
5.21
< 0.0001
< 0.0001
< 0.0001
0.0236
Supplementary table 2. Changes in timing of migration (in days year-1) for three phases of spring
migration (when the first 5, 50 and 95% of the spring total had been caught or observed each year)
in five long-distance passerine migrants when passing the Middle East (Eilat, Israel 29ºN 34ºE;
Tryjanowski and Yosef (2002)) and prior to breeding area arrival in northern Europe (Jurmo,
Finland 60ºN 22ºE; Vähätalo et al. 2004) during a 21 year period (1984 to 2004). Positive values
indicate delayed arrival while negative values indicate advanced arrival. Significant changes where
identified using linear regression analyses (* = P < 0.05). Number of years and number of birds (N)
included in the analyses from each site as well as the overall mean arrival date and mean migration
speed (km/day) for the 50% migration phase (mMS: the distance (3650 km) divided by the number
of days used between the Middle East and northern Europe) and percentages of second-year birds
(2.yr%) in the total number of birds trapped for the three species where ageing is possible are also
presented.
Species
Middle East
Lesser Whitethroat Sylvia curruca
Whitethroat Sylvia communis
Blackcap Sylvia atricapilla
Chiffchaff Phylloscopus collybita
Willow Warbler Phylloscopus trochilus
Northern Europe
Lesser Whitethroat Sylvia curruca
Whitethroat Sylvia communis
Blackcap Sylvia atricapilla
Chiffchaff Phylloscopus collybita
Willow Warbler Phylloscopus trochilus
Years
N Mean
mMS
2.yr%
5%
50%
95%
0.07
0.59 *
0.02
0.55
0.54
18
16
20
19
15
17,542
2,352
30,035
10,466
793
31-Mar
30-Mar
20-Apr
11-Mar
27-Apr
71
62
111
68
141
68
66
69
0.18
0.59
0.28
0.08
-0.20
0.52 *
0.63 *
0.50
0.67 *
-0.20
21
21
21
21
21
9,747
7,659
5,082
2,005
32,473
22-May
28-May
23-May
04-May
22-May
-
74
73
90
-0.27
-0.32
-0.49
-0.09
-0.20
-0.18
0.06
-0.21
-0.15
-0.18
0.11
0.26
0.02
0.45
-0.02
Supplementary methods
Detailed description of the data and the statistics used.
Ringing data
We included standardised ringing data from daily mist-netting of migratory songbirds collected
during spring migration in the period 1984 to 2004 at observatories located en route through the
Middle East (Eilat, Israel 29ºN 34ºE; Tryjanowski & Yosef (2002)) and prior to arrival at their
breeding areas in northern Europe (Jurmo, Finland 60ºN 22ºE; Ahola et al. (2004)). For Jurmo
observatory, we also included observations from standardised counts (see Vähätalo et al. 2004).
Only species where a minimum of 20 individuals were trapped or observed per season for a
minimum of 15 years at each locations were included. The five species of long-distance migrants
satisfying these criteria all winter in sub-Saharan Africa (see supplementary table 1 and figure 2).
Climate data
The normalised difference vegetation index (NDVI) offers a fully comparable measure of net
primary productivity and indirectly a measure of ecological conditions (e.g. food availability)
between regions and years (Pettorelli et al. 2005). NDVI is calculated as the normalised difference
in reflectance obtained by an Advanced Very High Resolution Radiometer (AVHRR) sensor of
NOAA polar orbiting satellites. The index is a measure of the degree of absorption by chlorophyll
in red wavelengths which is proportional to leaf chlorophyll density and provided by Clark Labs as
monthly NDVI maximum value composite taken from AVHRR data. Accordingly, we used
monthly NDVI maximum value composite for the period 1984 to 2000 as a proxy measure for the
actual environmental conditions in the wintering areas (after Wisz et al. 2007; see supplementary
figure 2 a-e) prior to onset of migration (African spring NDVI: mean NDVI values of February and
March) as well as in the areas traversed prior to arrival at the breeding areas (European spring
NDVI: mean NDVI values of April and May). Time periods are defined so they reflect the general
phenology of the species’.
Statistics
We tested the relationships presented in figure 2 for the two response variables (passage day and
migration duration) by fitting ‘Latitude’, ‘Year’, and ‘Species’ as fixed variables and ‘Migration
phase’ as a repeated factor in linear mixed models (proc MIXED; SAS 2000) We performed
backward elimination procedures on statistical models including all independent variables and their
interactions removing insignificant variables except when higher-level interaction was significant.
First, we investigated the effect of ‘Latitude’ (two location: Eilat, Middle East and Jurmo, northern
Europe), ‘Year’ (1984 – 2004), ‘Species’ (five species; see the text) and ‘Migration phase’ (first
5%, 50% and 95% of the total number of trapped or observed individuals) on timing of migration in
the Middle East and northern Europe. Arrival days were standardised to mean zero. Second, we
tested the effect of ’Latitude’, ‘Year’, ‘Species’ as well as African spring NDVI and European
spring NDVI on timing of migration in the Middle East and in northern Europe, respectively.
Arrival days and NDVI values were standardised to mean zero. Fitting a simple model including
only ‘Latitude’, ‘Year’ and ‘NDVI’ without interaction resulted in significant values of both year
and NDVI. Third, we analysed the effect of species, migration phase and European spring NDVI on
migration duration through Europe (number of days between the Middle East and northern Europe).
For each model, goodness of fit was confirmed by visual inspection of plots of residuals against
predicted values.
Supplemental references
Ahola, M., Laaksonen, T., Sippola, K., Eeva, T., Rainio, K. & Lehikoinen, E. 2004 Variation in
climate warming along the migration route uncouples arrival and breeding dates. Global Change
Biol. 10, 1610-1617.
Pettorelli, N., Vik, J. O., Mysterud, A., Gaillard, J. M., Tucker, C. J. & Stenseth, N. C. 2005 Using
the satellite-derived NDVI to assess ecological responses to environmental change. Trends Ecol.
Evol. 20, 503-510.
SAS 2000 SAS version 8.02. SAS Institute Inc., Cary, USA.
Tryjanowski, P. & Yosef, R. 2002 Differences between the spring and autumn migration of the redbacked Shrike Lanius collurio: record from the Eilat stopover (Israel). Acta Ornithologica 37, 8590.
Vähätalo, A. V., Rainio, K., Lehikoinen, A. & Lehikoinen, E. 2004 Spring arrival of birds depends
on the North Atlantic Oscillation. J. Avian Biol. 35, 210-216.
Wisz, M. S., Walther, B. A. & Rahbek, C. 2007 Using potential distributions to explore
determinants of Western Palaearctic migratory songbird species richness in sub-Saharan Africa. J.
Biogeography 34, 828-841.
Zink, G. 1973 Der Zug europäischer Singvögel vol. 1. Möggingen: Vogelzug-Verlag.