Supplementary material

Timing of migration, routes and wintering grounds of a short-distance diurnal migrant revealed by geolocation: A case study of Linnets Carduelis cannabina

Dennis Röseler*, Heiko Schmaljohann, Franz Bairlein

Institute of Avian Research “Vogelwarte Helgoland”, An der Vogelwarte 21, 26386 Wilhelmshaven

*for correspondence:

e-mail:

Data analysis

The geolocators sampled light intensity at 1 minute intervals and recorded the maximum light at every 2 minutes (MK6) or every 5 minutes (Intigeo). The measured light intensities of the retrieved tags were downloaded with the corresponding tools (MK6: BASTrak Suite; Intigeo: IntiProc) and were corrected for clock drift. Based on the recorded light intensity over time and the known time of sunrise and sunset, the longitude (time of midday) and latitude (day length) could be determined (Lisovski et al. 2012). Raw light-level geolocation data are available upon request at “Movebank” (www.movebank.org; project: ‘Helgoland Linnets’)

The threshold value were set to 3 (MK6 – BAS geolocator arbitrary scale) and 2 (Intigeo – lux scale). Measured light levels which crossed the threshold value were identified as twilight events for sunrise and sunset times. All daytime- and nighttime-transitions were manually checked for evident shading events, lightning events during the night and multiple and wrong sunrise and sunset events. After excluding outliners (20 ± 4% of the fixes, mean ± SD) from the six datasets of Linnet A, B, C1, C2, D and E they contained 426, 423, 534, 610, 431 and 706 fixes (one fix for noon and one for midnight) during 269, 267, 328, 370, 281 and 419 days. Twilight events of the remaining ones where the difference was >30 min from the otherwise stable surrounding were slightly adjusted.

Sun elevation for the breeding site could be calculated by using the pre-calibration- and the in-habitat-calibration-method, i.e. on-bird calibration, on the breeding area with the given coordinates of Helgoland. In order to calibrate the geolocators we used a pre-calibration- (open habitat) and in-habitat-calibration-method. To analyze the data of the devices the in-habitat-calibration method should be the first choice, because it shows the real light threshold level (Lisovski et al. 2012, Briedis et al. 2016). The calculated sun elevation angles may differ because of shading events and differences between the individuals. Sun elevation angles reached from - 1.4 to -2.1 (MK6; 2013) and -3.7 to -3.8 (Intigeo; 2014) during the times when birds were at the breeding area. The calculated mean error to Helgoland was 70 ± 29 km (mean ± SD, n = 590). In order to find appropriate sun elevation angles during the non-breeding period, we applied the Hill-Ekstrom calibration (range: -1.4 to -6.3) (Lisovski et al. 2012). Stationary periods during the non-breeding period were estimated with the “changeLight” function of the R package “GeoLight” (Lisovski & Hahn 2012). Stationary periods being not more than one day apart from each other and being nearly at the same position were merged. Points that fell over water were plotted to the nearest point on land. It should be clear that geolocation data always bear the risk of underlying an unknown error and cannot show the accuracy like GPS-Logger do (Rakhimberdiev et al. 2016).

Birds’ migration speed and great circle distance from the breeding area to the wintering ground did not correlate during autumn migration (Spearman rank correlation: S = 14, p-value = 0.2417, rho: 0.6). During spring migration we found a correlation between birds’ migration speed and the corresponding great circle distance (Spearman rank correlation: S = 4, p-value = 0.03333, rho: 0.88).

If not otherwise stated, mean ± SD are given.

Season 2013/2014

Season 2014/2015

Fig.A1: Longitude and latitude estimates (raw datasets) over time of the five tagged Linnets (6 datasets). The red lines indicate the longitude and latitude of the breeding area (Helgoland). Grey bars indicate time periods of 14 days before and 14 days after the equinoxes. Season 2013/2014: red: Linnet A, blue: Linnet B, green: Linnet C1. Season 2014/2015: green: Linnet C2, red: Linnet D, blue: Linnet E.

Geolocator compatibility

Geolocator mass of the MK6 represented <4.4% (mean: 3.8%) and of the Intigeo <5.4% (mean: 4.6%) of the birds body mass. The slightly higher load, than the 5% suggested upper permissible load limit (Cochran 1980), seemed unlikely to have adverse effects on the Linnets’ behaviour as in another songbird species (Naef-Daenzer et al. 2001). When checking recaptured birds in hand no injuries caused by the tags or leg-loop harness were observed. However, some tiny crumpled skin could be noticed where the geolocator was mounted. Mean body mass of not returned tagged Linnets was 18.4 g (± 1.4, range: 15.8 - 22.3 g, n = 52). The mean load of the light-level geolocator was 4.2%. Mean body mass of not-tagged Linnets was 18.1 g (± 1.1, n = 27) in 2013 and 18.3 g (± 1.8, n = 28) in 2014.

Annual cycle: Observed breeding birds

Almost daily observations on the island revealed that most Linnets arrive in the Helgoland breeding population between early April and end of May, with an average arrival at 2nd May ± 18 days (mean ± SD, n = 43). Males arrived slightly but insignificantly earlier than females (males: 29th April ± 15 days, n = 30; females: 9th May ± 21 days, n = 13; t-test, p = 0.142). Autumn departure commenced from August onwards, with females later than males (males: 5th September ± 15 days, n = 30; females: 15th September ± 13 days, n = 10; t-test, p = 0.076). The last breeding bird was observed on the island on 17th October.

Return rate

There was also no apparent difference in the return-rates of juveniles being tagged or not tagged but only colour-ringed (n = 4 of 37 vs 5 of 38; Chi2 = 0.077, p = 0.782) and of adults being tagged or not tagged (n = 5 of 24 vs. 2 of 29; Chi2 = 1.693, p = 0.193).

Tab.A1: Comparison between observed and geolocator data of dates and position estimates at the breeding area with given errors of the tagged birds. Birds are named as in Fig. 2.

Linnet A / Linnet B / Linnet C1 / Linnet C2 / Linnet D / Linnet E
sex / m / m / m / m / m / f
age / juv / ad / juv / ad / ad / ad
Body mass (in g) before tagging (mass without geolocator) / 18.7 / 21.9 / 19.5 / na / 18.5 / 18.6
Body mass (in g) after recapturing (mass without geolocator) / 20.7 / 21.3 / na / 20.4 / 19.0 / 18.2
Geolocator load (in %) / 3.7 / 3.2 / 3.6 / na / 4.6 / 4.6
observed departure date / 22.08.2013 / 14.08.2013 / 27.07.2013 / 19.09.2014 / 21.09.2014 / 17.07.2014
estimated departure date (based on geolocator data) / 23.08.2013 / 23.08.2013 / 07.10.2013 / 23.09.2014 / 06.10.2014 / 08.09.2014
difference in d / 1 / 9 / 41 / 4 / 15 / 53
observed arrival date / 15.04.2014 / 07.04.2014 / 27.04.2014 / 21.04.2015 / 20.04.2015 / 15.04.2015
estimated arrival date (based on geolocator data) / 12.04.2014 / na / 27.04.2014 / 19.04.2015 / 20.04.2015 / 19.04.2015
difference in d / 3 / na / 0 / 2 / 0 / 4
Logger period breeding area/in-habitat / 26.07.2013-22.08.2013 / 16.07.2013-14.08.2013 / 11.07.2013-27.07.2013 / 27.04.2014-19.09.2014 / 02.06.2014-21.09.2014 / 02.06.2014-17.07.2014
Logger down / 19.04.2014 / 11.04.2014 / 21.04.2015 / 24.04.2015 / 23.07.2014
estimated median position
latitude / 53.61 / 54.01 / 53.91 / 54.31 / 54.01 / 54.29
longitude / 8.55 / 8.89 / 8.89 / 8.53 / 8.40 / 7.19
known position
latitude / 54.18 / 54.18 / 54.18 / 54.18 / 54.18 / 54.18
longitude / 7.88 / 7.88 / 7.88 / 7.88 / 7.88 / 7.88
error distance km / 77.24 ± 38.66 / 68.67 ± 34.37 / 72.52 ± 36.31 / 44.42 ± 22.22 / 39.36 ± 19.69 / 46.39 ± 23.15

Tab.A2: Calculated main median stopover positions (given in latitude/longitude; [interquartile range]) and duration (in days [d]) of the tagged birds.

References

Briedis M, Beran V, Hahn S, Adamík P (2016) Annual cycle and migration strategies of a habitat specialist, the Tawny Pipit Anthus campestris, revealed by geolocators. J Ornithol 157: 619-626.

Cochran WW (1980) Wildlife telemetry. In Wildlife management techniques manual (ed. S. Schemnitz). 507-520. Washington, USA: The Wildlife Society.

Lisovski S, Hewson CM, Klaassen RHG, Korner-Nievergelt F, Kristensen MW, Hahn S (2012) Geolocation by light, accuracy and precision affected by environmental factors. Methods Ecol Evol: published online. (doi: 10.1111/j.2041-210X.2012.00185.x)

Naef-Daenzer B, Widmer F, Nuber M (2001) A test for effects of radio-tagging on survival and movements of small birds. Avian Sci 1: 15-23.

Rakhimberdiev E, Senner NR, Verhoeven MA, Winkler DW, Bouten W, Persma, T (2016) Comparing inferences of solar geolocation data against high-precision GPS data: annual movements of a double-tagged black-tailes godwit. Journal of Avian Biology, 47: 001-008.