Migrant bird populations often show substantial variation in route choice and timing. Determining whether this population-level variation is driven by between-individual differences and/or flexibility within individuals is key to identifying drivers of migration patterns. ‘Repeatability’ (R, the proportion of population-level variation attributable to between-individual variation) has become a central metric for the relative consistency of individual behaviour. Individual repeatability in migratory route choice and timing is often reported to vary between seasonal and regional contexts and may also differ between demographic groups (e.g. sexes), but interpreting repeatability requires careful consideration of the underlying changes in between- and within-individual variation. We GPS-tracked repeat migrations for eight male and five female Eleonora’s falcons Falco eleonorae and quantified the magnitude of within- and between-individual variation and the individual repeatability of their seasonal routes and timing at 100km intervals all across Africa. We did this across both sexes and then separately for males and females. We found greater between-individual variation in spring routes, albeit with substantial regional fluctuations in both seasons. The greatest between-individual variation in routes occurred during the spring desert-crossing, but this coincided with high within-individual variation, and thus only low repeatability of route choice. Route repeatability instead peaked (R = 0.6–0.8) through the Horn of Africa in spring and during the rainforest-crossing in autumn. Variation and repeatability of timing were stable across regions, with generally higher between-individual variation and repeatability in spring. Sex-specific analyses suggest males exhibit higher route repeatability, while females exhibit stronger seasonal contrasts in timing repeatability. Such sex differences were unexpected, but overall, between-individual variation and repeatability in routes and timings appear greater where environmental and annual cycle constraints are weaker. Route repeatability is especially high where falcons show fidelity to stop-over sites, or individual barrier-crossing preferences. Individual routines may be acquired through early-life exploration-refinement.

Tracking falcons

Since 2012 we have tagged 40 Eleonora’s falcons with UvA-BiTS GPS-trackers (7.5g, www.uva-bits.nl, Bouten et al. 2013) on Alegranza Islet (29°24'N, 13°30'W, 1050 ha, max 289 m a.s.l.). GPS-trackers were attached using a backpack-style harness made from 6.35 mm wide Teflon. Device and harness weighed ~8 g, corresponding to max. 2.63% and 2.42% of the mass at capture for males (min. 304g) and females (min. 330g), respectively. Data was downloaded locally through a custom-built antenna network, and up until the summer of 2020 we retrieved data for 19 falcons, and repeated tracks for 13 of these, Further details on return rates and apparent survival are provided in Vansteelant et al. (2021). All birds were sexed using molecular methods (see Gangoso & Figuerola, 2019, for further details).

Falcon trapping and tagging were approved by the Dirección General de Protección de la Naturaleza (Viceconsejería de Medio Ambiente), Canarian Government (permits n° 2014/2224, 2015/3835, 2017/6829, 2020/10521).

Processing migration data

Because tracking data were collected at variable time intervals (depending on e.g. dynamic measurement schemes, battery power) we resampled individual tracks to hourly intervals prior to further analyses. Next, we identified the start and end date of each migration trip based on a combination of daily movement statistics and geographical criteria. We first calculated the great-circle distance between the first to the last fix on each bird-day and classified days on which falcons covered a distance >100km as travel days. Days on which falcons covered shorter distances were considered as ‘stationary’ days. Tracks were then partitioned into ‘travel’ and ‘stationary’ segments, and we defined the end of autumn and the start of spring migration as the first and last day, respectively, on which falcons remained stationary for at least three days in Madagascar. Determining the start of autumn and the end of spring migration is more complicated because falcons spend up to two months at pre-breeding sites on mainland NW Africa before returning to the colony to breed. Because the focus of this study is on repeatability of directed long-distance movements, we considered that spring migration ended on the first day that falcons either reached the Atlantic Ocean or stopped-over in Morocco or the Western Sahara. Finally, we considered that autumn migration started on the day that falcons last visited the colony, thus including short stop-overs in NW Africa prior to the desert-crossing in the autumn migration.

Literature cited

Bouten, W., Baaij, E.W., Shamoun-Baranes, J. et al. 2013 A flexible GPS tracking system for studying bird behaviour at multiple scales. J Ornithol 154, 571–580. https://doi.org/10.1007/s10336-012-0908-1

Gangoso L, Figuerola J. 2019. Breeding success but not mate choice is phenotype- and context-dependent in a color polymorphic raptor. Behav Ecol 30(3):763-769. doi: 10.1093/beheco/arz013.

This tracking dataset can be processed using the R code provided at the GitHub repository of Wouter Vansteelant: https://github.com/Wouter-Vansteelant/Vansteelant-etal-2023-JAvianBiol

The repository contains all the code necessary to replicate our analyses and includes links to third-party online accessible datasets that were used for visualizing and analysing the data.