The migration patterns of birds breeding at high latitudes have undergone major changes during the Holocene, as species expanded from small refugia following the last glaciation. Unique features of genetic migration programs and species-specific dispersal patterns have resulted in various levels of migratory connectivity. High migratory connectivity can occur when populations expanding from different refugia maintain historically distinct wintering sites, or when species expanding their breeding ranges maintain their migratory vector, leading to corresponding shifts in the populations’ wintering areas. Alternatively, species may develop novel routes to nearby wintering sites during range expansion, also leading to high migratory connectivity. Here, we analyse geolocator and multisensor data logger tracks of great reed warblers Acrocephalus arundinaceus from a population at its eastern range limit in Kazakhstan. We compare their migration routes with published data from five Western Palearctic populations to understand how migration patterns have evolved as the breeding range expanded. Mitochondrial DNA data suggest that Kazakhstan was colonised from the western part of the range, but the logger data show that Kazakh birds winter in East Africa together with conspecifics from Turkey. This indicates that their migration route did not arise as a simple parallel shift of an unchanged vector-based programme but required drastic modifications of the migratory directions to maintain African wintering quarters. A remarkable finding in our study was the detection of a novel spring migration strategy. We found that birds leave the African winter quarter already in February to spend up to two months at an intermediate staging area in southern Iraq, half-way to their breeding grounds in Kazakhstan. We call this a two-step spring migration strategy and we outline three potential explanations for why such a behaviour may have evolved.

Description of the data and file structure

Light data from geolocators of seven tracked great reed warblers (BF574, BF578, BF583, BF590, BF591, BF593, BF597) tagged on breeding grounds in Kazakhstan (43.65°N, 76.82°E) in 2017 and retrieved 2018 in separate files (xxxxx.lux).

Activity data from multisensor loggers of two great reed warblers (XE32, XE41) tagged on breeding grounds in Kazakhstan (43.65°N, 76.82°E) in 2018 and retrieved 2019 in separate files (XXXX_acc.xlsx). Activity is stored per hour on the multsensor logger in six categories (acc[0] = no activity, acc[5] = max activity) for each of the 12 five minute blocks. The hourly activity (SumActivity) is calculated as the sum, taking the number for each category multiplied by its score value. For example, if all five minute blocks have acc[0], the activity is 0 x 12 = 12. If all five minute blocks have acc[5], the activity is 5 x 12 = 60.

Pressure data from a multisensor logger of one great reed warbler (XE32) tagged on breeding grounds in Kazakhstan (43.65°N, 76.82°E) in 2018 and retrieved 2019 in separate files (XE32_press.csv)