Spring migration passage advances can be counteracted by range shifts

Long-term passage records of migrating birds from fixed observatories are widely used to infer spring migration phenology. Yet migrants recorded at a site need not represent a constant mixture of populations over time. Spring onset and optimal arrival occur later at higher latitudes. If northbound birds pass en-route observatories later than southbound conspecifics, northward redistribution will increase their contribution to the passing pool and shift species-level passage relatively later—even if arrival advances within individual breeding populations. We tested this predicted spatial compositional bias. We related reported trends in the 10th, 50th and 90th percentiles of spring passage during 1980–2004 for 33 passerine migratory bird species at four Nordic observatories to north-northeast range-shift velocities. These velocities measured changes in the density-weighted centroids of Finnish breeding distributions from line-transect surveys. We used the Akaike information criterion to compare support for migration strategy, range-shift velocity, their additive and interactive effects, and an intercept-only baseline. Simulation extrapolation (SIMEX) accounted for species-specific measurement error in passage trends and range-shift velocities. Among short-distance migrants, relatively later 10th -percentile passage trends were associated with faster north-northeast redistribution. No comparable association was detected among long-distance migrants. After measurement-error correction, a 1 km year –1 faster north-northeast shift was associated with an early-passage trend 0.054 ± 0.022 days year –1 later, equivalent to approximately 1.3 days over 24 years. Because range shifts averaged close to zero among short-distance migrants in our study system, accounting for redistribution did not relevantly alter their group-average passage trend. At the species level, the contribution of the modelled range-shift effects could span several days and sometimes reverse the apparent trend. Median and late passage showed no clear support for the predicted positive association between north-northeast redistribution and relatively later passage. Fixed-site passage trends can be a result of phenological changes within breeding populations and redistribution among populations that migrate at different times. Here, evidence for the latter was restricted to early passage among short-distance migrants. Changes in breeding distributions should be considered when passage-date time series are used to infer arrival phenology at breeding grounds and assess climate-change responses.