Fig. 1

(a) Estimated non-breeding locations (n = 32) of 26 Nightingales fitted with geolocators in the UK between 2009 and 2017 (dark blue circles = primary sites, light blue circles = sites occupied at midwinter where different, repeated tracks for individuals indicated by centre colours), compared with six estimated non-breeding locations of four birds fitted with geolocators in The Gambia (red circles) at one site, KBO (see Methods) (yellow triangle). Bars show 95% confidence intervals of ___location estimates. See Supplementary Fig. 3a for locations without application of the land mask; (b) Estimated non-breeding locations for UK-tagged Nightingales (open circles) overlaid onto a habitat suitability map derived from an SDM based on independent survey data (see Methods), indicating their relative position within the non-breeding range. Two initial non-breeding locations from GPS tags are also shown (crosses) (See Methods); (c) Comparison of population dispersion at non-breeding grounds (mean distance between pairs of estimated locations, corrected for sample size) of Nightingales tagged with geolocators in the UK (n = 26), The Gambia (n = 4) and Ghana (n = 3). Bars show 84% confidence intervals, so non-overlap indicates significance at P < 0.05⁷. Population dispersion for UK birds is based on the first or sole wintering ___location. Known dispersion at African tagging locations was close to 0 and is indicated by a cross. Differences between known dispersion and observed dispersion can be attributed to error in light-level geolocation estimation of sites. Estimates are shown with and without the application of a land mask, demonstrating the impact that proximity to a coast has on dispersion estimates, which is considerable for birds tagged in the UK and Ghana (see Supplementary Fig. S3).