cross-hemisphere migration of a 25-gram...
TRANSCRIPT
1
Electronic Supplementary Material 1
2
Cross-hemisphere migration of a 25-gram songbird 3
4
Franz Bairlein, D. Ryan Norris, Rolf Nagel, Marc Bulte, Christian C. Voigt, 5
James W. Fox, David J. T. Hussell and Heiko Schmaljohann 6
7
Contents 8
1. Light-level geolocator data and analyses 9
1.1 Assessing the potential effect of geolocators on birds 10
1.2 Calibration of light-level geolocators 11
1.3 Estimation of the onset of migration (figure S1) 12
1.4 Estimation of migration routes (figures S2, S3) 13
1.5 Estimation of wintering grounds - outliers 14
15
2. Stable-hydrogen isotope analyses 16
2.1 Additional details on stable-hydrogen isotope analysis 17
2.2 Estimation of δD discrimination value 18
2.3 Assignment test: incorporating analytical and process error 19
2.4 Comparison of δD values between winter- and breeding-grown feathers 20
2.5 Comparison of δD values in wing coverts with geolocator data 21
2
1. Light-level geolocator data and analyses 22
1.1 Assessing the potential effect of geolocators on birds 23
Although tagging birds is invasive, the increase in flight costs are thought to be small [1], though drag 24
[2,3], as well as energy expenditure [4] may increase. To examine the potential effects of geolocators, 25
in 2008, we attached dummy geolocators, with the same weight and shape as real geolocators, to 12 26
northern wheatears housed in an indoor aviary at the Institute of Avian Research, Wilhelmshaven, 27
Germany. Over a six month period, we did not observe any differences in the flight behaviour of birds 28
compared to controls (without dummy tags), and found that treatment birds gained mass during the 29
migration period similar to controls, and as would be expected for birds in the wild. The leg-loop 30
harnesses fit irrespective of changes in the body mass and we did not observe any feather or skin 31
damage. 32
33
1.2 Calibration of light-level geolocators 34
Pre-deployment data were gathered in order to calibrate the devices. We set the light-level threshold 35
defining sunrise and sunset to 32 arbitrary data units [5] and, using the calibration data, this 36
corresponded to our adopted sun elevation angle of -3.5°. Accuracy of single fixes of three geolocators 37
to a reference site (54° 11’ N, 07° 55’ E; 20 May – 27 May 2009) in an open habitat during 38
precalibration was limited to 76 ± 33 km (n = 39). Here, light-level threshold was set again to 32 [5]. 39
Longitudinal deviation from the reference site was 45.6 ± 35 km and latitudinal deviation 47.1 ± 37.6 40
km (n fixes = 39). To illustrate the accuracy during precalibration independent of season and latitudinal 41
effects, we give the noon deviation (3.3 ± 2.1 min, n = 18) and midnight deviation (2.6 ± 2.2 min, n = 42
21). 43
44
1.3 Estimation of the onset of migration using geolocators 45
The program “TransEdit” was used to analyze the light data. All transitions were checked for obvious 46
shading events during the day and lighting events during the night and all “unnatural” transitions were 47
omitted [5,6]. Geolocators provide two fixes per day (one at noon and one at midnight). Because the 48
3
migration routes of both the Alaska and Canadian birds had a large longitudinal component, we 49
identified periods of migration based on longitude alone (see figure S1). 50
4
18.08.09
12.11.0912.11.09
AK-140° E
80° E
120° E
160° E
160° W
08.04.1008.04.10
40° E
80° E
120° E
160° E
160° W
AK-240° E
80° E
120° E
160° E
160° W
20.08.0920.08.09
10.11.0931.03.1031.03.10
23.05.1023.05.10
40° E
80° E
120° E
160° E
160° W
AK-340° E
80° E
120° E
160° E
160° W
18.08.09
01.12.0901.12.09
40° E
80° E
120° E
160° E
160° W
1.4.
15.4.
1.5.
15.5
29.03.1029.03.10
24.05.1024.05.10
CN-10° W
20° W
40° W
60° W
80° W
25.09.10
20.10.1020.10.10
15.8.
1.9.
15.9.
1.10.
15.10.
1.11.
15.11.
1.12.
Autumn migration
0° W
20° W
40° W
60° W
80° W
14.03.1114.03.11
8.05.118.05.11
1.3.
15.3.
1.4.
15.4.
1.5.
15.5
Spring migration
51
5
Figure S1. All raw (unsmoothed) longitudinal fixes (midnight and midday) over season for the 52
Alaskan (AK-1, AK-2, and AK-3; three upper) and the eastern Canadian (CN-1; lower panel) northern 53
wheatears. Black numbers indicate the dates corresponding to onset of autumn and spring migration 54
and end of autumn and spring migration. Plots on the left side describe autumn migration and on the 55
right side spring migration. Dotted black line indicates breeding site longitude. Longitudinal data for 56
bird AK-1 are shown until geolocator failure on 30 April 2010. AK-2 migrated for ten days and AK-3 57
for seven days north of the 70° latitude where, due to 24 hr daylight, no fixes were derived. Be aware 58
that (1) the x-axis (season) for spring migration and (2) the y-axis (longitude) differ between the 59
Alaskan and the Canadian birds. 60
61
1.4 Estimation of migration routes 62
a) Alaska (AK) 63
Generally, we derived the migratory pathways between the breeding area, stopover sites, and 64
wintering areas directly from the two fixes provided by light-level geolocators. We did not use 65
latitudinal data within a week on either side of the two equinoxes (22 September 2009, 20 March 66
2010), because latitude information is then usually unreliable [7]. In these cases, estimating whether a 67
bird arrived at or departed from a stopover site relied on longitudinal data, see also [5,6,8,9]. 68
For the AK birds, we did not consider latitude during autumn migration from 9 September – 69
16 October 2009 in AK-1, from 15 September – 18 October in AK-2, and from 13 September – 11
70
October 2009 in AK-3 (figure S2). These dates were selected based on the variation of the raw fixes 71
from the assumed migration route, see figure S2 for raw fixes and corresponding assumed migration 72
routes. As onset of spring migration occurred in all three cases well after spring equinox, we did not 73
have to omit any fixes for the estimation of spring migration route dues to such uncertainties. Cleaned 74
data set of birds AK-1, AK-2, and AK-3 contained 410, 486, and 483 fixes for the time periods 18 75
August 2009 – 30 April 2010, 20 August 2009 – 23 May 2010, and 20 August 2009 – 24 May 2010 76
(departure from the breeding area until arrival at breeding area). The low number of fixes from bird 77
AK-1’s geolocator was due to premature failure on 1 May 2010. The other two geolocators collected 78
light data until they were downloaded on 5 July 2010. 79
6
To determine a likely migration route between stopover sites, we smoothed latitude data where 80
possible by local polynomial regression loess fitting with raw longitude data as the predictor [10]. 81
Obvious latitudinal outliers (e.g. close to the equinoxes and due to light interferences) were omitted. 82
Because we used the raw longitude data to predict latitude using loess, we did not correct for variation 83
in longitude. Raw longitude data and the smoothed latitude estimations were used to indicate 84
migration routes (figure S2). 85
The R code for smoothing latitude estimation: A loess is calculated with a data set describing 86
the migration route between stopovers but excluding obvious outliers. We then used the predict 87
function of R by considering all data – including the outliers – to predict latitude estimations also for 88
the outliers. 89
90
Smoothed latitude = predict(loess(raw latitude data ~ raw longitude data, data excluding outliers, 91
span=1, degree=2), data including outliers) 92
93
To indicate the deviation of raw latitudinal data to each migration route given its longitude data, we 94
calculated the shortest (great circle) distances of each considered raw fix (excluding outliers) to the 95
estimated smoothed latitude of the corresponding longitude (figure S2). Figure S2 indicates that 96
precision of fixes varies strongly with season. Around equinoxes latitude cannot be reliably estimated. 97
Furthermore, topography and shading events may increase the inaccuracy of fixes further. By using a 98
smoothed line to indicate the migratory route, we likely underestimate the overall migration distance 99
and migratory speed because, in doing so, we did not consider small scale movements of the birds. 100
Migration distances were calculated using great circle distances [10] along the smooth lines between 101
stopover sites. 102
7
103
8
Figure S2. Interpolated geolocators tracks of two males (AK-1, AK-2) and one female (AK-3) 104
northern wheatears that bred in central Alaska, USA (65.6° N, 145.4° W; black dot). Blue = autumn 105
migration (mid August–mid November 2009), orange = spring migration (end March – mid May 106
2010). Dashed lines indicate rejection of data due to high uncertainty in latitude estimates (equinoxes 107
or light interference) or 24 hr daylight. Data logger of bird AK-1 failed during spring migration; the 108
last known fix is hypothetically joined along the shortest distance with the breeding area. Latitude data 109
were smoothed by local polynomial regression fitting with raw longitude data as the predictor (see text 110
for further information). In these cases, raw longitude data and the smoothed latitude estimations were 111
used to indicate the smoothed migratory routes between the corresponding stopover sites. Continuous 112
lines indicate these tracks. Accuracy of these tracks is estimated for each section as mean ± SD 113
distance (km) of the original fixes to the smoothed line and the number of considered fixes. Relevant 114
sections are indicated by bars and/or breeding/wintering area. Blue dots encircled black are all raw 115
fixes from autumn migration, orange dots encircled black are all raw fixes from spring migration. 116
117
b) Eastern Canadian Arctic (CN) 118
Because the CN bird started autumn migration (25 September 2010) close to the autumn equinox (23 119
September 2010), we could not estimate the actual migration route across the North Atlantic. After 120
reaching Western Europe, the bird basically migrated southwards. The initial movements after the 121
Atlantic crossing were still rather close to the autumn equinox (30 September 2010 and subsequent 122
days). Furthermore, as one can see in figure S3, the sequence of fixes did not show to a clear migration 123
route leading to the wintering grounds. Hence, we simplified the migration route as a straight line. The 124
onset of spring migration (14 March 2011) coincided with spring equinox (21 March 2011) so that 125
migration route from Africa to Western Europe was not possible to estimate. However, we could 126
estimate where the bird crossed the Atlantic with a higher degree of accuracy during the spring 127
because the crossing took place significantly after spring equinox and started most likely around 128
10.04.2011. Figure S3 shows the raw fixes during migration for the CN bird. 129
9
130
Latitude [°]
-10
0
10
20
30
40
50
60
70
80
90
Longitude [°]
-100 -80 -60 -40 -20 0 10 20 30 40 50
midnight 24/9 midnight 27/9
131
Figure S3. Interpolated geolocators tracks of one male (CN-1) northern wheatear that bred in eastern 132
Canada (63.7°N, 68.5°W; black dot). Blue = autumn migration (mid August – mid November 2009), 133
orange = spring migration (end March – mid May 2010). Dashed lines indicate rejection of data due to 134
high uncertainty in latitude estimations. Raw fixes of the Canadian bird during migration. Blue dots 135
encircled black are all raw fixes from autumn migration, orange dots encircled black are all raw fixes 136
from spring migration. Large blue dots are raw fixes from the autumn Atlantic crossing. 137
10
1.5 Estimation of wintering grounds - outliers 138
The estimation of the wintering grounds is explained in the main text. We excluded obvious outliers 139
(18 ± 3, mean ± s.d., range = 16 – 24 in bird AK-1; none in bird AK-2; 3 ± 2, range = 2 – 7 in bird 140
AK-3; and 14 ± 3, range = 13 – 21 in CN-1 with sun elevations ranging from -2° to -4.5° in steps of 141
0.5° for each bird) due to weather or bird behaviour, i.e. seeking shelter in dark hollows during day.142
11
2. Stable-hydrogen isotope analyses 143
2.1 Additional details on stable-hydrogen isotope analysis 144
Stable-hydrogen isotope values of feathers were analyzed at the stable isotope laboratory of the 145
Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany. A small section of feather tissue 146
was clipped from the tip of the feather (350 ± 7 µg) and loaded into silver capsules (IVA 147
Analysetechnik e.K. Meerbusch, Germany). We filled 96 port microtiter trays with silver capsules 148
loaded with feather samples and laboratory keratin standards with known δD values for non-149
exchangeable hydrogen. Trays were allowed to equilibrate with ambient air over more than seven 150
days. Afterwards, trays were placed in a drying oven over one day at 50°C. Then, loaded capsules 151
were transferred to a Zero Blank autosampler (Costech Analytical Technologies Inc., Cernusco, Italy) 152
above the elemental analyzer (HT elemental analyzer HEKAtech GmbH, Wegberg, Germany). For at 153
least 1 hour before combustion, samples were flushed in the autosampler with chemically pure helium 154
(Linde, Leuna, Germany). We used a Delta V Advantage isotope ratio mass spectrometer 155
(ThermoFischer Scientific, Bremen, Germany) that was connected via an interface (Finnigan Conflo 156
III, ThermoFisher Scientific, Bremen, Germany) with the elemental analyzer. Reference H2 gases were 157
calibrated against international standards (IAEA NBS 22 and IAEA-CH-7). We used the comparative 158
equilibration method [11] to account for the amount of exchangeable hydrogen in feather keratin. We 159
used three laboratory standards that covered the range of expected δD values in our samples. These 160
standards were also used to determine the δD of non-exchangeable hydrogen [11,12]. The stable 161
hydrogen isotope ratios of the non-exchangeable hydrogen (mean ± s.d.) of the standards were: -133.6 162
± 1.2‰, -109.1 ± 1.2‰ and -87.2 ± 1.0‰. In the sequential order of one autorun, keratin standards 163
were placed at positions 1-6 (3 standards of 2 replicates) and at the 9th-11
th position (3 standards). 164
165
2.2 Estimation of δD discrimination value 166
To determine how different growing season δDP values change the δD values of tail feathers 167
(summer), we sampled tail feathers from different breeding areas including eastern Canada (Iqaluit), 168
Iceland, ’marine‘ Scotland (Fair Isle), mainland Scotland, Morocco, Italy, western Germany 169
(Rhineland-Palatinate), Sweden (Uppsala), eastern Siberia (Provenja, Russia) and central Alaska 170
12
(Eagle Summit). The number of individuals sampled ranged from five to 30 northern wheatears per 171
breeding area. We then correlated the mean δD values of tail feathers (δDF) from the different 172
breeding areas with the growing season δDP values of the corresponding breeding areas (R2 = 0.94; n = 173
10; δDF = 0.662δDP - 24.3‰) [13,14]. All subsequently given δD values, also in figure 1, were 174
corrected using the regression equation for the discrimination factor between precipitation and feather 175
δD values. 176
177
2.3 Assignment test: incorporating analytical and process error 178
To incorporate analytical error, we re-sampled each δD value of wing covert 100 times, drawing from 179
a normal distribution where the mean was the δD value of the covert and the s.d. was taken from the 180
mean repeatability of the internal standard (± 1.9, n = 24). We incorporated process error from the 181
OIPC data by taking each of these values and calculating the likelihood assignment for each region 182
based on a randomly drawn value taken from a normal distribution with a mean and standard deviation 183
calculated from the sample of sites taken from the OIPC database. We considered a bird to have 184
originated from the region with the highest number of standardized probabilities out of the 100 re-185
samples. 186
187
2.4 Comparison of δD values between winter- and breeding-grown feathers 188
The stable-hydrogen isotope values of wing covert feathers that originated from the wintering quarters 189
differed significantly from the corresponding values of tail feathers moulted at the breeding area (AK 190
birds: winter coverts: δD: -27.3 ± 18.5‰, mean ± s.d., n = 9; breeding-ground tail feathers δD: -174.3 191
± 6.7‰, mean ± s.d., n = 29; ). Similarly, the δD values of fresh covert feathers of Baffin Island 192
breeding birds differed significantly from the δD values of tail feathers known to grow at the breeding 193
grounds only (winter area: δD: -40.4 ± 25.2‰, mean ± s.d., n = 7; breeding area: δD: -137.4 ± 12.2‰, 194
mean ± s.d., n = 102). 195
13
2.5 Comparison of δD values in wing coverts with geolocator data 196
Only one the birds returning to the AK with a geolocator had moulted a wing covert feather during the 197
previous winter in 2008/2009. Based on its δD value of this feather (-33.8‰), this individual was 198
predicted to have overwintered in central Sahelian Africa. Based on geolocator data, this bird appeared 199
to have overwintered in north-eastern Africa in 2009/2010. 200
201
202
References 203
1 Irvine, R. J., Leckie, F. M. & Redpath, S. M. 2007 The cost of carrying radio-transmitters: a test 204
with homing pigeons. Wildl. Biol. 13, 238–243. 205
2 Bowlin, M. S., Henningsson, P., Muijres, F. T., Vleugels, R. H. E., Liechti, F. & Hedenström, 206
A. 2010. The effects of geolocator drag and weight on the flight ranges of small migrants. 207
Methods Ecol. Evol. 1, 398–402. (doi: 10.1111/j.2041-210X.2010.00043.x) 208
3 Pennycuick, C. J., Fast, P. L. F., Ballerstädt, N. & Rattenborg, N. C. 2011 The effect of an 209
external transmitter on the drag coefficient of a bird's body, and hence on migration range, and 210
energy reserves after migration. J. Ornithol. published online. (doi: 10.1007/s10336-011-0781-211
3) 212
4 Barron, D. G., Brawn, J. D. & Weatherhead, P. J. 2010 Meta-analysis of transmitter effects on 213
avian behaviour and ecology. Methods Ecol. Evol. 1, 180-187. (doi: 10.1111/j.2041-214
210X.2010.00013.x) 215
5 Stutchbury, B. J. M., Tarof, S. A., Done, T., Gow, E., Kramer, P. M., Tautin, J., Fox, J. W. & 216
Afanasyev, V. 2009. Tracking Long-Distance Songbird Migration by Using Geolocators. 217
Science 323, 896. (doi: 10.1126/science.1166664) 218
6 Bächler, E., Hahn, S., Schaub, M., Arlettaz, R., Jenni, L., Fox, J. W., Afanasyev, V. & Liechti, 219
F. 2010 Year-Round Tracking of Small Trans-Saharan Migrants Using Light-Level Geolocators. 220
PlosOne 3, e9566. (doi :10.1371/journal.pone.0009566) 221
14
7 Fox, J. W. 2010 Geolocator manual. Version 8. Cambridge, UK: British Antarctic Survey. 222
8 Egevang, C., Stenhouse, I. J., Phillips, R. A., Petersen, A., Fox, J. W. & Silk, J. R. D. 2010 223
Tracking of Arctic terns Sterna paradisaea reveals longest animal migration. Proc. Natl. Acad. 224
Sci. 107, 2078–2081. (doi: 10.1073/pnas.0909493107) 225
9 Tøttrup, A. P. et al. 2011 The annual cycle of a trans-equatorail Eurasian-African passerine 226
migrant: different spatio-temporal strategies for autumn and spring migration. Proc. R. Soc. 227
Lond. B, published online (doi: 10.1098/rspb.2011.1323) 228
10 R Development Core Team 2010 R: a language and environment for statistical computing. 229
(Vienna: R Foundation for Statistical Computing. http://www.R-project.org). 230
11 Wassenaar, L. I. & Hobson, K. A. 2003 Comparative equilibration and online technique for 231
determination of non-exchangeable hydrogen of keratins for use in animal migration studies. 232
Isot. Environ. Health Stud. 39, 1–7. (doi: 10.1080=1025601031000096781) 233
12 Wassenaar, L. I. & Hobson, K. A. 2000 Improved method for determining the stable-hydrogen 234
isotopic composition of complex organic materials of environmental interest. Environ. Sci. 235
Technol. 34, 2354–2360. (doi: 10.1021/es990804i) 236
13 Hobson, K. A. 1999 Tracing origins and migration of wildlife using stable isotopes: a review. 237
Oecologia 120, 314–326. (doi: 10.1007/s004420050865) 238
14 Clark, R. G., Hobson, K. A. & Wassenaar, L. I. 2006 Geographic variation in the isotopic (δD, 239
δ13C, δ15N, δ34S) composition of feathers and claws from lesser scaup and northern pintail: 240
implications for studies of migratory connectivity. Can. J. Zool. 84, 1395–1401. (doi: 241
10.1139/z06-135) 242