Behavioral Ecology and Sociobiology

, Volume 66, Issue 6, pp 915–922

Tracking migration routes and the annual cycle of a trans-Sahara songbird migrant


    • Institute of Avian Research “Vogelwarte Helgoland”
  • Martin Buchmann
    • Unterer Sand 12
  • James W. Fox
    • Migrate Technology Ltd.
  • Franz Bairlein
    • Institute of Avian Research “Vogelwarte Helgoland”
Original Paper

DOI: 10.1007/s00265-012-1340-5

Cite this article as:
Schmaljohann, H., Buchmann, M., Fox, J.W. et al. Behav Ecol Sociobiol (2012) 66: 915. doi:10.1007/s00265-012-1340-5


Movement ecology studies have highlighted the importance of individual-based research. As tracking devices have not been applicable for identifying year-around movements of small birds until recently, migration routes of such species relied on visual observations and ring recoveries. Within the Palaearctic–African migration system, loop migration seems to be the overall migration pattern. The interindividual variations within species-specific migration routes are, however, unknown. Here, we track the individual migration routes and annual cycles of male Northern Wheatears Oenanthe oenanthe, a trans-Sahara songbird migrant from a German breeding population with light-level geolocators. Two migrated most likely via Spain towards western Africa but returned via Corsica/Sardinia, while two others seemed to migrate via Sardinia and Corsica in autumn and via Spain and France in spring (loop migration). The fifth took presumably the same route via France and the Balearics in both seasons. All birds wintered in the Sahel zone of western Africa. Overall migration distances for autumn and spring were similar (about 4,100 km), whereas the overall migratory speed was generally higher in spring (126 km day−1) than in autumn (88 km day−1). Birds spent about 130 days at the breeding area and 147 days at the wintering grounds.


Avian migrationLight-level geolocatorMigration routeNorthern WheatearSongbirdTrans-Sahara migrant


Bird migration has long fascinated mankind. Through simple observation, it is impossible to track birds from their breeding areas to their wintering grounds. The observation of the famous German “arrow stork” in 1822 proofed, finally, that birds breeding in Europe wintered in Africa because an African arrow stuck in the neck of the White Stork Ciconia ciconia on its return (Kinzelbach and Kinzelbach 2005). Nearly 200 years later, our knowledge about bird migration has increased dramatically by bird ringing, satellite and radio tracking, geolocation, radar and analysis of stable isotopes (reviewed in Robinson et al. 2009). In large species, migration patterns have been disentangled on an individual level, demonstrating the need for individual-based studies (Nathan et al. 2008), because variation between individuals of the same population might disguise individual patterns (Alerstam et al. 2006). In small birds, migration routes could be studied on species and population but not on an individual level (reviewed in Berthold 2001). Extrinsic devices were too heavy and detection probability of small ringed birds was too low to discover individual migration routes (Korner-Nievergelt et al. 2010). With intrinsic markers, one can only spot regions, in which a certain tissue had been growing (Hobson and Wassenaar 2008) but cannot track individual migration routes.

Within the Palaearctic–African migration system, loop migration seems to be a common pattern (Zink and Bairlein 1995; Berthold 2001), thereby birds take either a more westerly route in autumn than in spring (counterclockwise loop migration, Pied Flycatcher Ficedula hypoleuca, Red-backed Shrike Lanius collurio) or migrate the other way around (clockwise loop migration, Golden Oriole Oriolus oriolus; Winkel and Frantzen 1991; Bairlein 2001; Tøttrup et al. 2011). To determine individual migration routes and wintering grounds of a small trans-Sahara songbird migrant, we attached light-level geolocators (geolocators hereafter) to Northern Wheatears Oenanthe oenanthe (wheatears hereafter) of a breeding population in western Germany (Buchmann 2001). We tagged males because their return rate (56 %) is higher than that of females (43 %; Buchmann et al. 2009). Based on regularly occurring currents (Liechti 2006), we hypothesize that wheatears perform counterclockwise loop migration as many European songbirds do wintering in Africa (Berthold 2001). Although tens of thousands of wheatears were ringed in Europe, none was re-sighted in sub-Saharan Africa. Hence, we could not hypothesize a population-specific wintering ground within species’ known winter distribution (Walther et al. 2010). Further aims of this study were to determine the time devoted to the different life history stages, i.e., breeding, autumn migration, wintering and spring migration based on geolocator tracks.

Material and methods

Study site

We fitted stalked geolocators, Mk10S (1.2 g, with 13-mm sensor stalk at 30° angle to the horizontal), developed by the British Antarctic Survey (BAS) to 20 adult male wheatears breeding in a vineyard area (~15 km2) in Rhineland-Palatinate, western Germany (49°30′ N, 8°12′ E) in 2009 (Buchmann 2001). Body mass at capture was measured to the nearest 0.1 g.

Geolocator compatibility

Geolocators were fitted using Rappole–Tipton style harnesses (Rappole and Tipton 1990) that were made from elastic silicone–rubber mixture (MVQ Arcus,, and leg-loop length was adjusted individually (Naef-Daenzer 2007). The total attachment weighed 1.4 g. Because the lowest body mass of birds involved was 20.9 g (22.9 ± 1.3 g; mean ± SD, n = 20), tag mass represented <6.7 % (mean: 6.1 %) of the bird’s body mass. The relative load was, therefore, similar to but slightly higher than the suggested upper permissible load limits (Cochran 1980). Such deviation from the suggested 5 % limit did not seem to affect birds’ survival (Naef-Daenzer et al. 2001). Although tagging birds is invasive, no adverse effects were found when transmitters had been fitted appropriately (Naef-Daenzer et al. 2001), and the increase in flight costs is small (Irvine et al. 2007), though drag (Bowlin et al. 2010) as well as energy expenditure increase (Barron et al. 2010). To assess any direct effects of geolocators on wheatears before attaching them to wild birds, we attached dummy geolocators to 12 birds of our indoor breeding population for half a year in 2008. Birds could fly freely. They adjusted their body mass seasonally appropriated to high values during migration periods but low values otherwise. Leg-loop harnesses still fitted well, irrespective of these natural changes in body mass. Birds did not differ from control birds in their flight behavior and did not show damage of feathers or skin. The same was true for the free-flying tagged birds of the Rhineland-Palatinate population. We could find no apparent difference in 2010 in the clutch size between pairs with males carrying geolocators (5.8 ± 0.45 eggs; mean ± SD, n = 5) and pairs with males without geolocators (5.2 ± 1.0 eggs; mean ± SD, n = 74; Wilcoxon signed-rank test: W = 265, p = 0.089) and none in the first brood’s hatching success of males carrying geolocators (97 %, 28 chicks out of 29 eggs, n = 5) and of males without geolocators (91 %, 346 chicks out of 382 eggs, n = 74; \( \chi_1^2 = 0.0{13} \), p = 0.91; see, also electronic supplementary material). Future studies with a high sample size and lasting several years have to show whether there is a negative effect of geolocators on breeding performance and quality of their juveniles (Adams et al. 2009) which might be more pronounced in later years.

Data analysis

The BAS Mk10S geolocator measured light every minute and recorded the maximum value within each 10-min interval. Light data were processed with the programs provided by BAS (BASTrak suite). Programs were accomplished as given in Stutchbury et al. (2009). Light data were corrected for clock drift. The program “TransEdit” provided by BAS was used to analyze the light data and to identify the sun elevation corresponding with a particular light threshold level. All transitions were manually checked for obvious shading events during daytime and lighting events during nighttimes. As the wheatears inhabit open habitats during breeding, migration and winter (Cramp 1988), shading events are of minor importance in comparison to forest birds (Stutchbury et al. 2009; Fudickar et al. 2011). The “unnatural” transitions were omitted. This procedure was done similarly as, e.g., in Stutchbury et al. (2009) and Bächler et al. (2010).

In order to calibrate the devices, predeployment data was gathered in an open habitat (Helgoland, 54°11′ N, 07°55′ E) from the 13th to the 20th May 2009. We set the light-level threshold defining sunrise and sunset to 32 arbitrary data units as in Stutchbury et al. (2009). Using the calibration data (seven sunsets and seven sunrises per geolocator), this corresponded to sun elevation angles of −4.6 ± 0.8° for bird A, −4.7 ± 0.7° for bird B, −4.5 ± 0.9° for bird C, −4.5 ± 0.9° for bird D and −4.4 ± 0.9° for bird E (mean ± SD, n = 14). Single fixes of the five geolocators to the reference site (Helgoland) was limited to 128 ± 70 km (mean ± SD, n = 65). Longitudinal deviation from the reference site was 0.8 ± 0.5° corresponding to 51 ± 36 km (mean ± SD, n = 65) with given latitude of Helgoland, and latitudinal deviation was 0.9 ± 0.7° corresponding to 106 ± 78 km (mean ± SD, ns = 65) with given longitude of Helgoland. To illustrate the accuracy during precalibration independent of season and latitudinal effects, we give here the noon deviation (5.1 ± 2.9 min, mean ± SD, n = 30) and midnight deviation (5.6 ± 4.4 min, mean ± SD, n = 35).

When analyzing birds’ movements, we adjusted set sun elevation in such a way that deviation from the breeding area was smallest. This resulted in sun elevations of −4.7°, −4.7°, −4.0°, −4.2° and −4.0° for birds A to E which were similar to the ones retrieved during precalibration (see above). Centers of home ranges (50 % kernel densities) during the breeding period were 70 ± 54 km (mean ± SD, range: 15–135 km, n = 5) away from the actual breeding area; mean longitudinal deviation was 14 ± 4 km (mean ± SD, range: 9–19 km, n = 5) and mean latitudinal was 68 ± 56 km (mean ± SD, range: 7–133 km, n = 5) based on 96 fixes for bird A, 201 for bird B, 180 for bird C, 160 for bird D and 114 for bird E. The number of fixes differed between birds because geolocators were attached to the birds at different dates (22nd May–20th June 2009) and birds left the breeding area at different times of the year (Fig. 1a–e). As we could not test whether the selection of sun elevation for the breeding area applied equally well for the wintering grounds, we show 50 % kernel densities of the wintering areas for different sun elevations (−3°, −3.5°, −4°, −4.5°, −5° and −5.5°; see Fig. 1a–e). In doing so, we enlarged the area in which wheatears might have stayed over winter in 2009/2010. The covered areas will, however, most likely hold the true wintering grounds of the birds. Hence, we overcame the problem of applying a sun elevation derived from precalibration and the breeding area to times when birds were in Africa where local circumstance might lead to applying different sun elevations then at the precalibration and breeding areas.
Fig. 1

af Interpolated geolocator tracks of five male wheatears (a, b, c, d, e) that bred in Rhineland-Palatinate, western Germany (49°30′ N, 8°12′ E, black dot). Black indicates breeding area, blue indicates autumn migration, orange indicates spring migration and green indicates wintering ground. Density contour reflects 50 % kernel density of stopovers during migration and of wintering grounds. Dates in black are from longitude and latitude data, dates in grey are only estimated based on longitudinal data. Latitudinal data was not considered 10 days around the equinoxes. Question marks indicate uncertainties in the whereabouts of the birds during the times given next to the sign. Dotted lines between stopover sites indicate rejection of data due to uncertainty in latitude estimations (equinoxes or light interference). Continuous lines between stopover sites and/or breeding or wintering grounds indicate more accurate tracks. Kernel densities of wintering grounds estimated with the sun elevation derived from precalibration/the breeding area (green lines) and estimated with sun elevations ranging from −3° to −5.5° in steps of 0.5° (thin black lines) are indicated on the maps (see “Material and methods”). Averaged annual cycles of the five wheatears (f). Black indicates breeding area, blue indicates autumn migration, orange indicates spring migration and green indicates wintering grounds. Given are mean ± SD departures from and mean ± SD arrival dates at the breeding grounds and wintering grounds

BAS software’s simple movement compensation was applied to correct latitude for changes in day length caused by movements in longitude. The precision of longitude estimation is relatively constant throughout the year (Fig. 2; Fudickar et al. 2011), but in our case, the longitudinal component of the migratory route was rather small and latitude was most significant (Fig. S2); see, also Stutchbury et al. (2009) and Bächler et al. (2010). The accuracy of identifying stopover sites was, therefore, relatively low. We had to define a stopover where fixes were clustered for more than 2 days based on our own research with wheatears (Dierschke et al. 2005) and following Bächler et al. (2010). Because of the uncertainties when estimating birds’ whereabouts (about 130 km without influence of equinox), we did not apply a fixed maximum distance between adjacent fixes, though most were within 200 km. All raw fixes of the migration periods are shown in Fig. S1. Fixes during stopover were not allowed to be over water and were moved towards the nearest land except for the Mediterranean Sea. There, we could not exclude possible stopovers on islands. Hence, hypothetical stopover sites indicated around the Mediterranean might also cover the sea (Figs. 1a–e, S1). Stopovers and wintering grounds were defined by kernel densities encompassing 50 % of the maximum density. Kernel densities were calculated with the R package “adehabitat” using the ad hoc method for the estimation of the smoothing parameter and the bivariate normal kernel (R Development Core Team 2011). Because of the low accuracy of the fixes, we did not analyze the locations of the stopovers at a small spatial level. Hence, a higher precision than 50 % kernel densities of stopover sites was not necessary. We did not consider latitude data 10 days around each equinox (22nd September 2009, 20th March 2010). During that time, decisions whether a bird arrived at or departed from a stopover site relied on longitudinal data. Such data should be treated cautiously, since actual latitude of the bird remained unknown. It should be clear that the displayed migration routes include some unknown error especially close to the equinoxes. Uncertainties of migration routes and stopover sites were given in Fig. 1a–e. If latitude information was additionally missing due to light level interferences and birds’ behavior, departure and arrival events were estimated again from longitude data alone (see Figs. 1a–e, 2, S1). To demonstrate that, to a certain extent, birds’ movements can be estimated based on longitudinal data alone, we give here original longitudinal data and the corresponding smoothed lines (local polynomial regression fitting) over season for all birds (Fig. 2).
Fig. 2

Raw longitudinal data of the five wheatears over time of season. Given are raw longitude values per bird and their corresponding smoothed lines (local polynomial regression fitting; orange indicates bird A, blue indicates bird B, dark green indicates bird C, black indicates bird D, grey indicates bird E). The periods of 10 days around each equinox are highlighted in light grey. The dashed black line indicates longitude of the breeding area (8.2°E)

We derived the migratory pathways between the breeding area, stopover sites and wintering grounds directly from the fixes (Fig. S1). We defined birds’ migration distance between the breeding area, centers of stopover sites and wintering grounds as the minimum distance, i.e., along great circle routes. Overall average migratory speed was calculated as bird’s migration distance over its minimum migratory duration, i.e., onset of migration until arrival at breeding or wintering grounds. This underestimates the overall migration distance and migratory speed rather than overestimating these values because small-scale movements (<150 km) and deviation from the assumed migration route cannot be considered when applying geolocators to birds with a strong latitudinal migration component close to the equinoxes. Distances and speed estimations are, therefore, conservative estimates. Even average values across individuals should be interpreted cautiously due to the small sample size.

After excluding obvious outliers close to equinoxes, due to heavy clouds or bird behavior, i.e., seeking shelter in dark hollows during the day (25 ± 5 % of the fixes; mean ± SD, n = 5), cleaned datasets contained 491, 590, 607, 494 and 495 fixes (one fix for noon and one for midnight) during 318, 366, 412, 355 and 328 days. One geolocator (bird A) had a premature failure and collected light data after its arrival in the breeding area only until the 4th of April 2010. It was seen at the breeding area on the 2nd of April 2010 and identified by its color rings. The other geolocators collected light data until download on the 14th, 21st and 23 rd of May 2010 and the 10th of July 2010.

Calibration after deployment was gathered in an open habitat (power station, Wilhelmshaven, 53°36′ N, 08°06′ E) from the 26th to 30th June 2010 for the four still running geolocators. The corresponding sun elevation angles were −4.8 ± 0.2° for bird B, −4.7 ± 0.7° for bird C, −4.7 ± 1.0° for bird D and −4.6 ± 1.0° for bird E (mean ± SD, n = 16, eight sunsets and eight sunrises). Single fixes to the reference site (Wilhelmshaven) was limited to 128 ± 70 km (mean ± SD, n = 65). Longitudinal deviation from the reference site was 1.8 ± 1.5° corresponding to 123 ± 69 km (mean ± SD, n = 54) with given latitude of Wilhelmshaven, and latitudinal deviation was 0.2 ± 0.1° corresponding to 22 ± 14 km (mean ± SD, n = 54) with given longitude of Wilhelmshaven. Longitude estimations depend on local noon with respect to a known clock. If the internal geolocator clock drifts over time, the longitude coordinates drift correspondingly, whereas latitude coordinates are hardly affected. For the correction of clock drift, a linear drift over time is assumed. However, if clock does not drift linearly, there is a bias in the longitude estimates (Fox 2010). In our case, there was a bias in three geolocators towards the east of about 2° at the end of the study (Fig. 2). A nonlinear clock drift explains most likely the apparent mismatch of longitudinal data during the breeding season in 2010 in comparison to the actual longitude of the breeding area (Fig. 2). As we cannot identify when this bias in longitude occurred, centers of stopover sites and wintering grounds might slightly be too far east (Fig. 1a–e). A difference of 2° longitude at the equator equals a distance of about 220 km and at the breeding area of about 140 km. Such a bias does not, however, change the overall migration routes of the birds. It should be generally clear that migration routes, stopover sites, breeding areas and wintering grounds identified by geolocator data always bear an unknown error.


Effect of light-level geolocators on birds

The return rate of wheatears tagged with geolocators (45 %, nine out of 20) was not different from the one of only ringed males (54 %, 58 out of 107; \( \chi_1^2 = 0.0{88} \), p = 0.766). One bird had lost its geolocator before the first sighting in 2010. Another one was most likely killed after its arrival in April 2010. Two others had lost their geolocator before recapture but after arrival in the breeding area in 2010. Light-level data was successfully downloaded from five devices.

Migration route

Birds A and B initiated autumn migration through central Europe in a southerly direction, most likely crossing the Mediterranean Sea via Corsica and Sardinia. They performed lengthy stopovers (about 2 and 3 weeks) in western Algeria before reaching their wintering grounds in western Mali and Mauritania. In spring, migration both stopped over for about 2 weeks in March, most likely in Spain or France, before arriving at the breeding area at the end of March and mid-April (Fig. 1a, b). Birds C, D and E first followed a southwesterly direction and reached eastern Spain. Because the crossing of the desert coincided with autumn equinox, indicated migratory pathways and stopovers are vague. Birds wintered in Burkina Faso, Mauritania and Mali. Most of their spring migration occurred around the equinox so that given migration routes are uncertain (Fig. 1c–e). However, there is evidence that bird C crossed the Mediterranean Sea via the Balearic Islands, whereas birds D and E most likely did so via Sardinia and Corsica possibly indicating clockwise loop migration (Figs. 1c–e, S1).

The maximum distance between the centers of birds’ wintering grounds was about 1,500 km (bird B–bird C) regardless of the sun elevation applied for the estimation of the fixes (Fig. 1a–e). Applying higher sun elevations than those derived from calibration shifted the wintering grounds further to the south, and in applying lower sun elevations, potential wintering grounds were located further to the north.

Migration distance and speed

The average migration distance was similar in autumn (4,040 ± 240 km; mean ± SD) and spring (4,218 ± 273 km; mean ± SD; paired Wilcoxon signed-rank test: V = 13, p = 0.19, n = 5). Overall migratory speed was 88 ± 26 km day−1 (mean ± SD) during autumn and 126 ± 30 km day−1 (mean ± SD) during spring (paired Wilcoxon signed-rank test: V = 9, p = 0.81, n = 5). These values are rough estimates (see above and below).

Annual cycle

Wheatears left their breeding area around mid-August and arrived after about 48 days at the wintering grounds where they stayed for about 147 days. Arrival at the wintering grounds was difficult to ascertain because it happened close to autumn equinox. Onset of spring migration was around the beginning of March. Spring migration (about 36 days) was generally shorter than autumn migration, though not statistically significant (paired Wilcoxon signed-rank test: V = 23, p = 0.31, n = 5; Fig. 1f). Wheatears arrived at the breeding area in early April. Assuming birds left the breeding area as in the previous year, wheatears had about 130 days for breeding and moult before the onset of autumn migration (Fig. 1f). Interindividual variation in the onset of autumn and spring migrations and variation in the arrival dates at the wintering grounds and breeding area did not differ significantly (Friedman test: \( \chi_3^2 = {6}.{913} \), p = 0.075).


Effect of light-level geolocators on birds

Geolocators were about 6.1 % of birds’ body mass at capture and, therefore, higher than the commonly used upper permissible load of 5 % (Cochran 1980). During breeding, wheatears store only little fat, but on migration, birds increase their body mass by about 40–100 % of their “lean body mass” (Dierschke et al. 2005). Hence, wheatears might be adapted to carrying high fuel loads. Since return rate of geolocator-tagged wheatears was similar to species general site fidelity (50–56 %; Conder 1989; Buchmann et al. 2009) and since birds carrying geolocators behaved normally at the breeding area, we are confident that geolocators had little negative effect on male wheatears.

Migration route

If the predominating wind systems in Europe and Africa have formed loop migration (Liechti 2006), we would expect migration routes to be consistent within similar species. In raptors, however, loop migration is present in Marsh Harriers Circus aeruginosus (Klaassen et al. 2010) but absent in Montagu’s Harriers Circus pygarus (Trierweiler et al. 2007). The intraspecific variation in the migration routes of wheatears was similar to the one demonstrated for four hoopoes Upupa epops being tracked from their Swiss breeding area to their sub-Saharan wintering grounds and back (Bächler et al. 2010). Hoopoes have a similar bounding flight style as songbirds and can be regarded as a near passerine trans-Saharan migrant. Both studies did not support a population-specific migration route, but sample size was low and migration around equinoxes leads to spatial uncertainties. Studies with higher sample sizes have to determine whether there is a certain dominating migration route in these and similar species possibly driven by predominating wind systems (Liechti 2006).

Wintering grounds

Tracked wintering grounds matched species’ modeled winter distribution within the Sahel and northern savannah zone of sub-Saharan Africa (Walther et al. 2010) but might be slightly too far north in birds A, B, D and E (Fig. 1a, b, d, e). This discrepancy could be explained by a lack of observations in this region (Salewski et al. 2005) or by an artefact because precalibration of the geolocators took place at a higher latitude and in summer, leading towards more northern fixes in winter. Hence, the northern parts of the estimated wintering grounds (lower sun elevations) are unlikely to cover wheatears’ true wintering grounds.

Total migration speed

Individual values of migration speed must be treated cautiously because: (1) migration took place around the equinoxes, (2) small-scale movements (<150 km) cannot be considered with geolocators, (3) only one midday and one midnight fix are available per day not allowing tracking of the actual nocturnal flight path and (4) low sample size. Nevertheless, the seasonal averages of wheatear’s total migration speed seem to reflect the general pattern with higher values in spring (126 km day−1) than in autumn (88 km day−1; compare Hall-Karlsson and Fransson 2008; Stutchbury et al. 2009; Tøttrup et al. 2011). Speed estimates coincided well with the one of hoopoes (autumn: 81 km day−1, spring: 122 km day−1; Bächler et al. 2010). In Alaskan wheatears, the total migration speed was nearly twice as high (autumn: 160 km day−1, spring: 250 km day−1; Bairlein et al. 2012). The apparent difference in the total migration speeds between both wheatear populations is explained by the longer distance of Alaskan birds (about 14,500 km vs. 4,100 km) covered in less than twice the time birds of the German population have to accomplish their migration (Bairlein et al. 2012). Hence, German wheatears seemed to be less time constraint to complete the life history stages within the given time than Alaskan birds.

Annual cycle

Former knowledge of the different life history stages were based on visual observations of different individuals (Conder 1989; Panov 2005). Here, we show the annual cycle of five individually year-around tracked male wheatears (Fig. 1f). The average time spent at this breeding area was about 20 days shorter than previously assumed (Buchmann 2001). This difference is most likely explained by bird A leaving the breeding area prior to its postbreeding moult which starts usually during the first half of July and is accomplished before autumn migration (Buchmann et al. 2009). Excluding bird A, time spent at the breeding area (137 ± 7.5 days; mean ± SD, n = 4) was similar as determined by field observations (about 140 days; Buchmann 2001) indicating that this estimate based on geolocator data and a low sample size is similar to what we observe in the field.

In contrast to the German population, central Asian wheatears breeding in mountain tundra have less time for breeding and postbreeding moult. Adults in heavy primary moult were observed feeding their chicks in Tien Shan (Panov 2005). By overlapping these two life history stages (own observations), they optimized the temporal flexibility in the timing of autumn migration (Wingfield 2008). This rarely happens in the German population and only if birds perform a late second brood (Buchmann et al. 2009).

Use of light-level geolocation for bird migration research

Weight of geolocators was currently reduced to 0.6 g (Swiss Ornithological Institute and Biotrack Ltd., UK) enabling to track theoretically most bird species (≥12 g). Despite this weight advantage, there are several shortcomings of geolocation (see above). Birds must be recaptured. The maximum number of geolocators to be regained is defined by the species’ natural return rate, if geolocators do not affect this rate. However, the final number of total tracks is usually lower because birds might lose their geolocators before recapture, especially when using Rappole–Tipton style harnesses, or because of device failures (e.g., this study, Bächler et al. 2010; Tøttrup et al. 2011). Additionally to these problems, in restricting the sample size, there are several methodological limitations. (1) The general accuracy of location fixes is about 100–150 km (this study, Stutchbury et al. 2009; Fudickar et al. 2011) and (2) is limited by high latitudinal uncertainty during equinoxes. (3) The main migration period of many European and North American songbirds coincide with equinoxes (this study; Stutchbury et al. 2009; Bächler et al. 2010; Tøttrup et al. 2011). Latitudinal movements are, therefore, less convenient to investigate than longitudinal movements (Fudickar et al. 2011). This is also true because uncertainties in light attenuation (e.g., clouds) typically cause smaller errors in longitude than latitude, regardless of season (Ekstrom 2004; Lisovski et al. 2012). (4) Small-scale movements (<150 km) should not be considered, because the general accuracy of location fixes is in this order of magnitude (this study; Fudickar et al. 2011). (5) If calibration is usually carried out only at the breeding area, it is advisable to apply different sun elevations for the estimation of the winter fixes to cover most likely the true wintering ground (Lisovski et al. 2012). (6) Bird’s behavior (Daan and Aschoff 1975) and/or (7) bird’s environment change over season (Lisovski et al. 2012). E.g., preference for or occurrence of more dense vegetation on the wintering grounds than at the breeding area might lead to a later experienced sunrise and earlier experienced sunset (Fudickar et al. 2011). Nevertheless, geolocator studies have revolutionized our understanding of bird movements (e.g., Stutchbury et al. 2009; Bächler et al. 2010; Tøttrup et al. 2011; Bairlein et al. 2012).


This work was supported financially by the Deutsche Forschungsgemeinschaft (BA 816/15-4). We thank Ommo Hüppop and Freimut Schramm for technical support and Lesley Szosteck for improving the English. Four anonymous reviewers had greatly improved former versions of the manuscript.

Ethical standards

Northern Wheatears were caught, ringed and tagged under license of the Ministry for Agriculture, the Environment and Rural Areas, Rhineland Palatinate, Germany.

Supplementary material

265_2012_1340_MOESM1_ESM.pdf (303 kb)
ESM 1(PDF 303 kb)

Copyright information

© Springer-Verlag 2012